CN101837165B - Walking aid electrostimulation fine control method based on genetic-ant colony fusion fuzzy controller - Google Patents

Abstract

The invention relates to the field of rehabilitation training. In order to optimize the quantization factor, proportional factor and fuzzy control rules of the fuzzy controller, and then accurately and stably control the current mode of the FES system in real time, and can effectively improve the accuracy and stability of the FES system, the technical scheme adopted in the present invention is: based on The precise control method of walking-assisting electrical stimulation based on genetic ant colony fusion fuzzy controller includes the following steps: first, the selection of fuzzy control decision variables is transformed into a combinatorial optimization problem applicable to genetic and ant colony algorithms, and the decision variables are coded and randomly generated Chromosomes composed of n individuals; secondly, use the genetic algorithm to generate the initial pheromone distribution in the ant colony algorithm, and use the ant random search to optimize the membership function, quantization factor and proportional factor of the fuzzy controller; repeat the self-learning and self-adjusting process according to the aforementioned output , and ultimately used in the FES system. The present invention is mainly applied to rehabilitation training.

Description

Precise control method of electrical stimulation for walking aid based on genetic ant colony fusion fuzzy controller

technical field

The invention relates to the field of rehabilitation training, in particular to the measurement of the load-carrying force of a walker, and in particular to a precise control method for electric stimulation of walking aids based on a genetic ant colony fusion fuzzy controller.

Background technique

Functional Electrical Stimulation (FES) is a technology that stimulates limb motor muscles and peripheral nerves through current pulse sequences to effectively restore or reconstruct part of the motor function of paraplegic patients. At present, due to the weak regeneration ability of the spinal cord, there is no effective treatment method that can directly repair the injury for paralyzed patients with spinal cord injury. The implementation of functional rehabilitation training is an effective measure. According to statistics, the number of paralyzed patients with spinal cord injury is increasing year by year, and functional rehabilitation training is an urgently needed technology. In the 1960s, Liberson successfully used electrical stimulation of the peroneal nerve for the first time to correct the gait of hemiplegic patients with foot drop, and created a new way of functional electrical stimulation for motor and sensory function rehabilitation. At present, FES has become an important rehabilitation treatment method for restoring or reconstructing part of the motor function of paraplegic patients. However, how to precisely control the trigger timing and pulse current intensity of FES to ensure that the effect of electrical stimulation can accurately complete the predetermined functional action is still the key to the technology of FES. According to statistics, there is still little research on the trigger control of FES at present, and according to the deviation between the effect and the predetermined action, the closed-loop control is used to automatically adjust the FES stimulation intensity and timing parameters, thereby greatly improving the real-time performance, accuracy and stability of the FES system. However, effective control methods are still being explored.

Fuzzy controller is a method of formalizing and modeling human experience through fuzzy logic and approximate reasoning, and turns it into a control model acceptable to computers, allowing computers to replace people in real-time control of advanced strategies and control of the controlled object. The novel technical method can effectively improve the controllability, adaptability and rationality of the control algorithm, especially for the complex and difficult to model with mathematical equations and has a unique advantage in the subject of rich manual control experience, while the human muscle The complexity and time-varying operating environment make it possible to establish its mathematical model, which makes it difficult for traditional control methods to adapt to the strict requirements of the FES field. The fuzzy controller provides a new solution for the precise control of FES. The core technology of the fuzzy controller is to determine the structure of the fuzzy controller, the fuzzy rules used, the synthetic reasoning algorithm and the method of fuzzy decision-making. factors and fuzzy control rules and other parameter tuning. In the field of FES, the requirements for system stability are extremely strict, so the selection of fuzzy controller parameters is also particularly important. The method proposed in this patent based on the genetic algorithm to tune the fuzzy controller to control the precise control of functional electrical stimulation can achieve good results in the current level of the FES system.

The genetic algorithm belongs to the artificial biological evolutionary algorithm. It is a kind of natural selection and natural genetic mechanism for reference in the biological world. The main feature is the group search strategy and the information exchange between individuals in the group. Optimizing nonlinear problems. However, there is nothing that can be done about the utilization of feedback information in the system, resulting in a large number of powerless redundant iterations when the solution reaches a certain range, and the efficiency of obtaining an accurate solution is low. The ant colony algorithm simulates the foraging behavior of ants in the biological world looking for the shortest path from the ant nest to the food source without any prompts. A population-based simulated evolutionary algorithm is proposed, which has strong adaptability, distributed parallel computing, and is easy for others The advantages of algorithm integration, but the lack of pheromone in the initial stage, the solution speed is slow. Both genetic algorithm and ant colony algorithm have the advantages of wide application range and strong versatility, and are widely used in discrete optimization problems. Genetic algorithm and ant colony algorithm are complementary and organically integrated, which can overcome their respective shortcomings and give full play to their respective advantages .

Contents of the invention

In order to overcome the deficiencies in the prior art, the object of the present invention is to propose a new precision control method for FES, through the integration of genetic algorithm and ant colony algorithm, the advantages are complementary, and the quantization factor, scale factor and fuzzy control factor of fuzzy controller are optimized. Control rules, and then accurately and stably control the current mode of the FES system in real time, and can effectively improve the accuracy and stability of the FES system, and obtain considerable social and economic benefits. In order to achieve the above-mentioned purpose, the technical solution adopted in the present invention is: a precise control method for electric stimulation of walking aid based on genetic ant colony fusion fuzzy controller, comprising the following steps:

Firstly, the selection of 12 decision variables kfuzz of fuzzy control quantization factor, proportional factor and membership function parameter is transformed into a combinatorial optimization problem applicable to genetic and ant colony algorithm, and these 12 kfuzz are binary coded, and then n are randomly generated The initial population P(0) composed of individuals, where kfuzz is a vector of n×12;

Secondly, establish a reasonable objective function of the corresponding relationship between the actual joint angle and the output joint angle of the muscle model and determine the parameter settings of the ant colony algorithm, use the genetic algorithm to generate the initial pheromone distribution in the ant colony algorithm, and use the ant random search to optimize the fuzzy controller Membership function, quantization factor and proportional factor, and call the fuzzy controller that has been tuned to verify whether the preset goal is reached. If not, repeat the above operations until the parameters converge or reach the predetermined index, and output the decision variables of the fuzzy control and the operation of the ant colony the number of times;

According to the above-mentioned output fuzzy control decision variables, calculate the system output and its deviation from the muscle model under the fuzzy controller, then enter the next step of self-learning and self-adjustment, repeat this process, and finally realize the adaptive online tuning of fuzzy controller parameters , and used in the FES system.

Described fuzzy controller is a two-dimensional fuzzy controller, and two input variables actually output the error e (k) of joint angle and desired joint degree and the rate of change ec (k) of error respectively, and the domain of discussion is FE=[-E, E], FEC=[-EC, EC], the output stimulation current intensity u(k), its discourse domain is FU=[-U, U];

The domain of quantification of error is X={-n,-n+1,...0,...,n-1,n}; (1)

The domain of quantification of error change rate is X1={-m,-m+1,...0,...,m-1,m}; (2)

The domain of quantification of the control quantity is Y={-k,-k+1,...0,...,k-1,k}; (3)

The quantization factors are

K _e = n/X _e ; (4)

K _ec =m/X _ec ; (5)

The scale factor is

K _u = k/Y _u ; (6)

The domain of discourse of error: {-3 -2 -1 0 1 2 3}; the domain of discourse of error rate of change is {-3 -2 -1 0 1 2 3} the domain of discourse of output value {-3 -2 -1 0 1 2 3}. The control rules are: if E1 and EC1, then U1, if E2 and EC2, then U2, ... Ep and ECp, then Up;

Its total fuzzy control rule is

R＝(E _i ×CE _i ) ^T1 οC _i (8)

where E ₁ =(a _1i ...a _ni ), EC ₁ =(b _1i ...b _mi ), U ₁ =(c _1i ...c _ti )(i=1, ...p)

The defuzzification method used is the weighted average method

${\mathrm{<span\; class="notranslate">\; u</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}}{\overset{\mathrm{<span\; class="notranslate">\; the\; s</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; i</span>}{\mathrm{<span\; class="notranslate">\; k</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; /</span><span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}}{\overset{\mathrm{<span\; class="notranslate">\; the\; s</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; u</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 9</span>}<span\; class="notranslate">\; )</span>$

For each specific observation value deviation E ^* and its error rate of change EC ^* , then use their respective quantification factor formulas to become elements in the quantitative domain, and then fuzzify them into E ^* and EC ^* ,

where E ^* = (e ₁ ... e _n ), EC ^* = (f ₁ ... f _m )

Calculate the exact amount of output by formula (8).

The selection of the 12 decision variables of quantization factor, proportional factor and membership function parameter of the fuzzy control is transformed into a combinatorial optimization problem applicable to the genetic and ant colony algorithm, which is carried out according to the following formula:

K _e1 =kfuzzi(1)*K _e (18)

K _c1 =kfuzzi(2)*K _c (19)

K _u1 = kfuzzi(3)*K _u (20)

The adjustment of the membership function of the error is the adjustment of the domain of the error discourse as follows:

{-3-kfuzzi(4), -2-kfuzzi(5), -1-kfuzzi(6), 0, 1+kfuzzi(6), 2+kfuzzi(5), 3+kfuzzi(4)}

The setting of the membership function of the rate of error change is the setting of the universe of the error rate of change as follows:

{-3-kfuzzi(7), -2-kfuzzi(8), -1-kfuzzi(9), 0, 1+kfuzzi(9), 2+kfuzzi(8), 3+kfuzzi(7)}

The setting of the membership function of the output current is the setting of the domain of the membership function of the output current:

{-3-kfuzzi(10), -2-kfuzzi(11), -1-kfuzzi(12), 0, 1+kfuzzi(12), 2+kfuzzi(11), 3+kfuzzi(10)}

The initial pheromone distribution in the ant colony algorithm is generated by using the genetic algorithm, and the membership function, the quantization factor and the scaling factor of the fuzzy controller are optimized by ant random search. The aforementioned kfuzzi is not only the coding of the chromosome in the genetic algorithm, but also the The code of the city, the length of the two codes should be equal, according to

n _i <log ₂ [(x _imax -x _imin )×10 ^m ]-1(21)

n _i ≥ log ₂ [(x _imax -x _imin )×10 ^m +1](22)

kfuzzi encoding length

The real number corresponding to each parameter can be decoded, and the corresponding decoding formula is

$\mathrm{<span\; class="notranslate">\; kfuzzi</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; min</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; l</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; 2</span>}}^{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>\frac{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; max</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; min</span>}}}{{\mathrm{<span\; class="notranslate">\; 2</span>}}^{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; twenty\; three</span>}<span\; class="notranslate">\; )</span>$

Among them, kfuzzi(i) is the variation of the fuzzy controller, l _i is the coded length of this parameter, b∈[0,1], x _imax and x _imin are the maximum and minimum values of the decision-making quantity respectively.

The specific steps are detailed as follows:

The first is the definition and setting of the genetic algorithm:

Step1: Initialize the genetic algorithm control parameters (population size, hybridization probability, mutation probability);

Step2: Set the end conditions of the genetic algorithm including: the minimum number of genetic iterations, the maximum number of genetic iterations, the minimum evolution rate, and the number of consecutive iterations;

Step3: Perform binary coding on the problem to be solved, and randomly initialize the population X(0)=(x ₁ , x ₂ , . . . x _n ).

Step4: For each individual _xi in the current group X(t), decode and calculate its fitness F(xi ₎ .

Step5: Determine the probability of each individual according to the individual fitness value and the roulette selection strategy, and directly inherit the two individuals with high probability to the next generation, and perform crossover and mutation operations according to the crossover and mutation algorithm.

Step6: Update the fitness value of the new generation of individuals. Repeat Step5.

Step7: Select individuals with strong adaptability and put them into the set as the optimal solution set. For each individual in the optimization set, the result of the genetic algorithm is set as the initial pheromone of the ant colony algorithm.

Again, the improvement of the ant colony algorithm in the genetic ant colony fusion algorithm, and then the genetic algorithm is connected, and this fusion algorithm is used to tune the parameters of the fuzzy controller:

Step8: Initial value setting of pheromone: τ _S =τ _C +τ _G , where τ _C is a constant, that is, the minimum pheromone in the MMAS algorithm, and τ _G is the pheromone value converted from the solution result of the genetic algorithm;

Step9: Parameter initialization: set time t=0 and number of cycles N _max =0, set the maximum number of cycles N _cmax , and place m ants at the starting point. Set the number of ants and the number of cycles;

Step10: Ants search randomly. After a crawl is over, decide which feature variables are selected as the actual input variables, and modify the taboo table pointer, that is, move the ant to a new element after selection, and move the element to the taboo of the ant individual table;

Step11: Calculate the fitness function and membership function, the probability selection elements calculated by the ant individual state transition probability formula;

Step12: Use the information provided by the training samples to generate the conditions of the fuzzy rules, and check the accuracy of the fuzzy model;

Step13: If the ant element has not been traversed, go to Step10, otherwise go to Step12;

Step14: Update the pheromone concentration of the characteristic variable with high classification accuracy of pheromone concentration to be enhanced, and will be selected with a greater probability in the next search;

Step15: End adjustment when satisfied, and end of setting.

The present invention is characterized in that: first, the current mode of functional electrical stimulation is controlled in real time according to the change of knee joint angle by fuzzy control, which effectively solves the difficulty in precisely controlling the current mode of electrical stimulation due to the complexity and time-varying nature of human muscles Second, using the good optimization characteristics of the genetic ant colony algorithm to adjust the quantization factor, scaling factor and membership function parameters of the fuzzy controller in real time, and then more accurately control the current mode of functional electrical stimulation, which can complete the ideal stimulation effect, and can prevent fatigue, the functional electrical stimulation system can be widely promoted, and obtain considerable social and economic benefits.

Description of drawings

Fig. 1 Convergence speed graph of genetic ant colony algorithm.

Fig.2 Structure diagram of fuzzy controller parameters tuning by genetic ant colony fusion algorithm

Figure 3 Experimental scene diagram.

Fig. 4 The fuzzy controller tracking result diagram of adaptive optimization tuning of genetic ant colony algorithm.

Fig. 5 The relative error between the preset input joint angle and the actual output under the control of fuzzy controller parameters by genetic ant colony algorithm tuning.

Detailed ways

The structure of the application of the precise control method of walking aid functional electrical stimulation based on genetic ant colony fusion fuzzy control is shown in Figure 2. Its working process is as follows: firstly, the selection of 12 decision variables of quantization factor, proportional factor and membership function parameter of fuzzy control is transformed into a combinatorial optimization problem applicable to genetic and ant colony algorithm, and it is coded and n individuals are randomly generated The chromosomes (initial population) formed, and then establish a reasonable relationship between the actual joint angle and the muscle model output joint angle and determine the parameter settings of the ant colony algorithm, making full use of the rapidity, randomness, and global convergence of the genetic algorithm. The result is to generate the initial pheromone distribution in the ant colony algorithm, use ants to randomly search for variables to optimize the membership function, quantization factor and proportional factor of the fuzzy controller, and call the fuzzy controller that has been tuned to verify whether the preset goal is achieved , if not, repeat the above operations until the parameters converge or reach the predetermined target; the final output is the decision variable of the fuzzy control and the number of times the ant colony runs. After calculating the system output and its deviation with the muscle model under the new fuzzy controller, enter the next step of self-learning and self-adjustment. Repeat this process, and finally realize the adaptive online tuning of fuzzy controller parameters, and use it in FES system.

1 Fuzzy control design

Due to the particularity of human beings, the FES field has strict requirements on controller stability, robustness, and real-time performance. The two-dimensional fuzzy controller is selected for the balance stability and real-time performance of the fuzzy controller, that is, the two input variables are actually output respectively. The error e(k) of the joint angle and the expected joint degree and the rate of change ec(k) of the error, the domain of discussion is FE=[-E, E], FEC=[-EC, EC], the output stimulation current intensity u (k), its domain of discourse is FU=[-U, U].

The domain of quantification of error is

X={-n,-n+1,...0,...,n-1,n}(1)

The domain of quantification of error rate of change is

X ₁ ={-m,-m+1,...0,...,m-1,m}; (2)

The domain of quantification of the control quantity is

Y={-k,-k+1,...0,...,k-1,k} (3)

The quantization factors are

K _e =n/X _e (4)

K _ec =m/X _ec (5)

The scale factor is

K _u =k/Y _u (6)

This patent adopts the domain of discourse of error: {-3 -2 -1 0 1 2 3}; the domain of discourse of error change rate is {-3 -2 -1 0 1 2 3} the domain of discourse of output value {-3 - 2 -1 0 1 2 3}. The control rules are: if E1 and EC1, then U1, if E2 and EC2, then U2, ... Ep and ECp, then Up;

Its total fuzzy control rule is

R＝(E _i ×CE _i ) ^T1 οC _i (8)

where E ₁ =(a _1i ...a _ni ), EC ₁ =(b _1i ...b _mi ), U ₁ =(c _1i ...c _ti )(i=1, ...p)

The defuzzification method used is the weighted average method

(9)

(9)

For each specific observation value deviation E ^* and its error rate of change EC ^* , then use their respective quantification factor formulas to become elements in the quantitative domain, and then fuzzify them into E ^* and EC ^* ,

where E ^* = (e ₁ ... e _n ), EC ^* = (f ₁ ... f _m )

There is Equation 8 to find the exact amount of output.

The angle error of this patent and its error rate of change are on [-90 90], and the domain of discussion is [-3 3], then the formula can be used

$\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>\frac{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; b</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; a</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; x</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; a</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; b</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 11</span>}<span\; class="notranslate">\; )</span>$

5.2 Genetic and ant colony fusion algorithm tuning fuzzy controller parameters

Ant colony algorithm is a new type of bionic algorithm derived from the biological world of nature. When using ant colony algorithm to solve optimization problems, the optimization problem is first transformed into the shortest path problem. Each ant starts from the initial node N ₀₀ or N ₀₁ , and walks sequentially through one of the child nodes of N ₁ , N ₂ ... until the end points N _k0 and N _k1 form a path (N _0t N _1t ... N _kt ), t ∈ [0, 1]. Its path can represent a binary feasible solution. Each time an ant visits a city, it has the following characteristics:

State transition rule: The state transition rule used by the ant colony algorithm is a random proportional rule based on the TSP problem, which gives the probability that ant k located in city i chooses to move to city j,

Among them, τ _ij (i, j) is the fitness of (i, j), and η _ij (i, j) is the reciprocal of the distance. α is the relative importance of residual information, and β is the relative importance of expected value.

In the ant colony algorithm, the selection method is

Among them, q is a random number uniformly distributed on [0, 1], and q ₀ is a parameter variable on [0, 1].

Global update rules: The ant algorithm has different update algorithms. The global update principle adopted by the ant colony system only allows the ants with the global optimal solution to release pheromone. neighborhood of the best path.

τ _ij (i, j)←(1-ρ)□τ _ij (i, j)+ρ·Δτ _ij (i, j)(14)

Among them, ρ is the volatility coefficient of information number, and L _gb is the local update information of the global optimal path found so far: each ant also updates the information number trace in the process of establishing a solution

τ _ij (i, j)←(1-γ)τ _ij (i, j)+γ·Δτ _ij (i, j)(16)

where γ ∈ [0, 1].

When using functional electrical stimulation to stimulate muscles to complete corresponding actions, the current should not be overexcited or the muscles should complete the action quickly. If the current exceeds the human threshold, you will feel pain, sometimes unbearable, and easily cause muscle fatigue, so It is especially important to quickly tune the relevant parameters of the fuzzy controller. Therefore, the advantages of genetic algorithm and ant colony algorithm are combined to improve the optimization efficiency and quickly adjust the relevant parameters of the fuzzy controller. The use of genetic ant colony fuzzy controller means the use of genetic ant colony fusion algorithm to adjust the distribution of quantization factor, proportional factor and membership function in fuzzy rules. as shown in the following formula

K _e1 =kfuzzi(1)*K _e (18)

K _c1 =kfuzzi(2)*K _c (19)

K _u1 = kfuzzi(3)*K _u (20)

The adjustment of the membership function of the error is the adjustment of the domain of the error discourse as

{-3-kfuzzi(4), -2-kfuzzi(5), -1-kfuzzi(6), 0, 1+kfuzzi(6), 2+kfuzzi(5), 3+kfuzzi(4)}

The setting of the membership function of the rate of error change is the setting of the universe of the rate of error change as

{-3-kfuzzi(7), -2-kfuzzi(8), -1-kfuzzi(9), 0, 1+kfuzzi(9), 2+kfuzzi(8), 3+kfuzzi(7)}

The setting of the membership function of the output current is to set the domain of the membership function of the output current as

{-3-kfuzzi(10), -2-kfuzzi(11), -1-kfuzzi(12), 0, 1+kfuzzi(12), 2+kfuzzi(11), 3+kfuzzi(10)}

Among them, kfuzzi is not only the code of the chromosome in the genetic algorithm, but also the code of the city in the ant colony algorithm. In order to make the two better integrated, the binary code is used, and the kfuzzi is unchanged in the fusion algorithm, so the length of the two codes should be equal. should follow

n _i <log ₂ [(x _imax -x _imin )×10 ^m ]-1(21)

n _i ≥ log ₂ [(x _imax -x _imin )×10 ^m +1](22)

kfuzzi encoding length

The real number corresponding to each parameter can be decoded, and the corresponding decoding formula is

$\mathrm{<span\; class="notranslate">\; kfuzzi</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; min</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; l</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; 2</span>}}^{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>\frac{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; max</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; min</span>}}}{{\mathrm{<span\; class="notranslate">\; 2</span>}}^{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; twenty\; three</span>}<span\; class="notranslate">\; )</span>$

Among them, kfuzzi(i) is the variation of the fuzzy controller, l _i is the coded length of this parameter, b∈[0,1], x _imax and x _imin are the maximum and minimum values of the decision-making quantity respectively.

The first is the definition and setting of the genetic algorithm:

Step1: Initialize the genetic algorithm control parameters (population size, hybridization probability, mutation probability).

Step2: Set the end conditions of the genetic algorithm (minimum number of genetic iterations, maximum number of genetic iterations, minimum evolution rate, number of consecutive iterations).

Step3: Perform binary coding on the problem to be solved, and randomly initialize the population X(0)=(x ₁ , x ₂ , . . . x _n ).

Step4: For each individual _xi in the current group X(t), decode and calculate its fitness F(xi ₎ .

Step5: Determine the probability of each individual according to the individual fitness value and the roulette selection strategy, and directly inherit the two individuals with high probability to the next generation, and perform crossover and mutation operations according to the crossover and mutation algorithm.

Step6: Update the fitness value of the new generation of individuals. Repeat Step5.

Step7: Select individuals with strong adaptability and put them into the set as the optimal solution set. For each individual in the optimization set, the result of the genetic algorithm is set as the initial pheromone of the ant colony algorithm.

Again, the improvement of the ant colony algorithm in the genetic ant colony fusion algorithm, and then the genetic algorithm is connected, and this fusion algorithm is used to tune the parameters of the fuzzy controller:

Step8: Initial value setting of pheromone: τ _S =τ _C +τ _G , where τ _C is a constant, that is, the minimum pheromone in the ant colony MMAS algorithm, and τ _G is the pheromone value converted from the genetic algorithm solution result.

Step9: Parameter initialization: set time t=0 and number of cycles N _max =0, set the maximum number of cycles N _cmax , and place m ants at the starting point. Set the number of ants and the number of cycles.

Step10: Ants search randomly. After a crawl is over, decide which feature variables are selected as the actual input variables, modify the taboo table pointer, that is, move the ant to a new element after selection, and move the element to the taboo of the ant individual table.

Step11: Calculate the fitness function and membership function, the probability selection element calculated by the ant individual state transition probability formula.

Step12: Use the information provided by the training samples to generate the conditions of the fuzzy rules, and check the accuracy of the fuzzy model.

Step13: If the ant element has not been traversed, go to Step10, otherwise go to Step12

Step14: Update the pheromone concentration of the characteristic variable with a high accuracy rate of pheromone concentration is enhanced, and will be selected with a greater probability in the next search.

Step15: Finish the adjustment when it is satisfied, and the setting is over.

Among them, genetics and ant colony share the fitness function, and the muscle characteristics are particularly complex and human tolerance is limited. This requires the three indicators of the control system to be measured as stability, accuracy and rapidity. In order to make the control deviation tend to zero, there is Faster response speed and smaller overshoot, so the fitness function should take the relationship between the feedback deviation e(t) between the optimal prediction and the expected value, the rate of change of the deviation ec(t) and the control variable u(t), The optimal index J selected as a parameter

$\mathrm{<span\; class="notranslate">\; J</span>}<span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; \&Integral;</span>}_{\mathrm{<span\; class="notranslate">\; 0</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}{\mathrm{<span\; class="notranslate">\; w</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; e</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; w</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; u</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; w</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; ec</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; dt</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 14</span>}<span\; class="notranslate">\; )</span>$

Wherein, w ₁ , w ₂ and w ₃ are weight values, and generally take w ₁ =100, w ₂ =10, and w ₃ =1.

The fitness function is

F=C/J (15)

Among them, C=10 ⁿ (n is an integer), when the individual fitness differs greatly, n≤0; when the difference is small, n≥0.

2 Experimental scheme

The experimental device used the Parastep functional electrical stimulation walking aid system produced by SIGMEDICS Company of the United States, the PS-2137 protractor and Data Studio software of PASCO Company. The Parastep system includes a microprocessor and a stimulation pulse generation circuit, including six stimulation channels, and is powered by a battery. The content of the experiment is: use the FES system to stimulate the relevant muscle groups of the lower limbs, use the PS-2137 protractor of PASCO Company to collect the knee joint angle and record the measured knee joint angle with Data Studio software. The subjects are required to be in good health, without lower extremity muscle and bone diseases, without neurological diseases and severe cardiopulmonary diseases. During the experiment, the subject sat on the test bench, fixed the stimulating electrodes at the two ends of the quadriceps, fixed the protractor on the thigh and calf, and made the joint movement point close to the knee joint movement point. When no electrical stimulation was applied, the calf was relaxed and kept in a vertical suspension state. The FES experimental scene is shown in Figure 3. The electrical stimulation pulse sequence adopts the classic Lilly waveform, the pulse frequency is 25Hz, the pulse width is 150μs, and the pulse current is adjustable within the range of 0-120m. In the experiment, the stimulation intensity can be adjusted by changing the magnitude of the pulse current to change the knee joint angle generated by the stimulation. Before the experiment, set the expected trajectory of the knee joint angle, and use the goniometer to detect the change of the knee joint opening angle in real time during the experiment. The sampling rate of the experimental data is 128Hz, and the data recording time is 60s.

Beneficial effect

The new fuzzy controller parameter algorithm of genetic ant colony fusion is used to measure and adjust the amplitude of FES pulse current, so that the knee joint angular motion generated by FES is close to the expected motion track. Figure 4 shows the tracking results of the fuzzy controller for the adaptive optimization of the genetic ant colony algorithm. The red line in the figure represents the expected motion trajectory, and the blue line represents the actual output joint angle. The X-axis is time, and the Y-axis is the knee joint motion angle. In order to observe the control error of the fuzzy controller set by this algorithm more clearly, as shown in Figure 5, the relative error between the preset input knee joint angle and the actual knee joint angle under the fuzzy control set by the genetic ant colony fusion algorithm, it can be seen that the error average Within 3%, precise control can be achieved.

The gist of the present invention is to propose a new precise control method for FES, which uses the genetic ant colony fusion algorithm to set fuzzy controller parameters, and then accurately, stably, and effectively controls the current intensity of the FES system in real time. The invention can effectively improve the real-time performance, accuracy and stability of the FES system, and obtain considerable social and economic benefits. The best implementation plan is to use patent transfer, technical cooperation or product development.

Claims (3)

1. walk help electro photoluminescence precision control method that merges fuzzy controller based on genetic-ant colony; It is characterized in that; Comprise the following steps: that at first selection with 12 decision variable kfuzzi of quantizing factor, scale factor and the subordinate function parameter of fuzzy control is converted into the combinatorial optimization problem that heredity and ant group algorithm are suitable for; And these 12 kfuzzi are carried out binary coding, and produce the initial population P (0) that the n individuals is formed afterwards at random, wherein kfuzzi is the vector of n * 12; The selection of 12 decision variables of said quantizing factor with fuzzy control, scale factor and subordinate function parameter is converted into heredity and the combinatorial optimization problem that ant group algorithm is suitable for, and is to be undertaken by following formula:

K _e1＝kfuzzi(1)*K _e (18)

K _c1＝kfuzzi(2)*K _c (19)

K _u1＝kfuzzi(3)*K _u (20)

To the subordinate function of error adjust promptly to the domain of error domain adjust into:

{-3-kfuzzi(4)，-2-kfuzzi(5)，-1-kfuzzi(6)，0，1+kfuzzi(6)，2+kfuzzi(5)，3+kfuzzi(4)}

To the subordinate function of error rate adjust promptly to the domain of error rate adjust into:

{-3-kfuzzi(7)，-2-kfuzzi(8)，-1-kfuzzi(9)，0，1+kfuzzi(9)，2+kfuzzi(8)，3+kfuzzi(7)}

The subordinate function of output current adjust promptly to the domain of the subordinate function of output current adjust into:

{-3-kfuzzi(10)，-2-kfuzzi(11)，-1-kfuzzi(12)，0，1+kfuzzi(12)，2+kfuzzi(11)，3+kfuzzi(10)}

The said genetic algorithm of utilizing produces plain distribution of initial information in the ant group algorithm; Utilize the ant random search to optimize subordinate function and the quantizing factor and the scale factor of fuzzy controller; Aforementioned kfuzzi is a chromosomal coding in the genetic algorithm; Be again the coding in city in the ant group algorithm, the length of the two coding should equate, according to

n _i＜log ₂[(x _imax-x _imin)×10 ^m]-1(21)

n _i≥log ₂[(x _imax-x _imin)×10 ^m+1](22)

Kfuzzi code length

The pairing real number of each parameter can be decoded and obtained, and then corresponding decoding formula does

$\mathrm{kfuzzi}\left(i\right)=x_{i\min }+(\underset{j=0}{\overset{l_i}{\mathrm{\&Sigma;}}}{b}_j\&times;2^{j-1})\&times;\frac{x_{i\max }-x_{i\min }}{2^{l-1}}---\left(23\right)$

Wherein kfuzzi (i) is the variable quantity of the fuzzy controller adjusted, l _iThe length of the coding of parameter for this reason, b ∈ [0,1], x _ImaxAnd x _IminBe respectively the maximal value and the minimum value of decision content;

Next sets up rational actual joint angles and the corresponding relation objective function of muscle model output joint angles and the parameter setting of definite ant group algorithm; Utilize genetic algorithm to produce plain distribution of initial information in the ant group algorithm; Utilize the ant random search to optimize subordinate function and the quantizing factor and the scale factor of fuzzy controller; And call the fuzzy controller of having adjusted, whether checking reaches goal-selling, if the above operation of no repetition; Perhaps reach predetermined index up to parameter convergence, the number of times of the decision variable of output fuzzy control and ant crowd operation;

According to the decision variable of aforementioned output fuzzy control under fuzzy controller computing system output and with the deviation of muscle model after get into next step again self study and self-adjusting; This process repeatedly; The final self-adaptation on-line tuning of realizing Fuzzy Controller Parameters, and be used for the FES system.

2. a kind of walk help electro photoluminescence precision control method based on genetic-ant colony fusion fuzzy controller according to claim 1 is characterized in that said fuzzy controller is a two-dimensional fuzzy controller; Two input variables are respectively the error e (k) and the error change rate ec (k) of actual output joint angles and expectation joint degree, and domain is FE=[E, E]; FEC=[EC, EC], the stimulating current intensity u (k) of output; Its domain is FU=[U, U];

The quantification domain of error is X={-n ,-n+1 ... 0 ..., n-1, n}; (1)

The quantification domain of error rate is X ₁=m ,-m+1 ... 0 ..., m-1, m}; (2)

The quantification domain of controlled quentity controlled variable is Y={-k ,-k+1 ... 0 ..., k-1, k}; (3)

Quantizing factor is respectively

K _e＝n/X _e；(4)

K _ec＝m/X _ec；(5)

Scale factor does

K _u＝k/Y _u；(6)

The domain of error: { 3-2-1 012 3}; The domain of error rate be the domain of 3-2-1 012 3} output valves 3-2-1 012 3}, control law is: if E1 and EC1 then U1, if E2 and EC2 then U2 ... Ep and ECp be Up then;

Its total fuzzy control rule does

R＝(E _i×EC _i) ^T1οC _i(8)

E wherein ₁=(a _1iA _Ni), EC ₁=(b _1iB _Mi), U ₁(c _1iC _Ti) (i=1 ... P)

The reverse gelatinization method that adopts is a method of weighted mean

For each concrete observed value deviation E ^*With its error rate EC ^*, use quantizing factor formula separately to become the element that quantizes in the domain more respectively, again its fuzzy E that turns to ^*And EC ^*,

E wherein ^*=(e ₁E _n), EC ^*=(f ₁F _m)

Ask the exact value of output by formula (8) and formula (9).

3. a kind of walk help electro photoluminescence precision control method based on genetic-ant colony fusion fuzzy controller according to claim 1 is characterized in that said step is refined as:

At first be the definition and the setting of genetic algorithm:

Step1: initialization genetic algorithm controlled variable: population scale, hybridization probability, variation probability;

Step2: the genetic algorithm termination condition is set comprises: minimum genetic iteration number of times, maximum genetic iteration number of times, minimum evolution rate, subsequent iteration number of times;

Step3: treating deals with problems carries out binary coding, random initializtion population X (0)=(x ₁, x ₂... x _n);

Step4: to each individual x among the current X of colony (t) _i, its fitness F (x is calculated in decoding _i);

Step5: confirm the probability that each is individual according to individual fitness and gambling wheel selection strategy, the direct heredity of two individuals that probability is high is arrived of future generation, and algorithm is carried out intersection, mutation operation according to intersecting, making a variation;

Step6: upgrade the individual adaptive value of a new generation, repeat Step5;

Step7: select the strong individuality of adaptive faculty, put into set, as the optimum solution set, to optimizing each concentrated individuality, it is plain that genetic algorithm result is set to the ant group algorithm initial information;

The improvement of ant group algorithm in the genetic-ant colony blending algorithm once more, genetic algorithm is connected mutually then, and with this blending algorithm Fuzzy Controller Parameters of adjusting:

Step8: the initial value setting of pheromones: τ _S=τ _C+ τ _G, wherein, τ _CBe a constant, promptly the minimal information in the MMAS algorithm is plain, τ _GIt is the pheromones value that the genetic algorithm for solving result transforms;

Step9: parameter initialization: make time t=0 and cycle index N _Max=0, maximum cycle N is set _Cmax, m ant placed starting point, ant number and cycle index are set;

Step10: ant random search; After the end of once creeping, determine the actual input variable of the selected conduct of which characteristic variable, revise the taboo list index; After promptly choosing ant is moved to new element, and move to this element in the individual taboo table of ant;

Step11: calculate fitness function and subordinate function, the probability that ant individual state transition probability formula calculates is selected element;

Step12: utilize the condition of the information generating fuzzy rule that training sample provides, the accuracy of check fuzzy model;

Step13:, change Step10, otherwise be Step14 if the ant element has not traveled through;

Step14: the pheromone concentration that the plain concentration of lastest imformation is divided the high characteristic variable of accuracy is enhanced, and next time can be selected with bigger probability when searching for;

Step15: satisfy termination condition, the end of adjusting.

Images