A platform elevator artificial intelligence control system and an artificial intelligence control method

Abstract

The application provides a platform elevator artificial intelligence control system and an artificial intelligence control method based on the platform elevator artificial intelligence control system. The platform elevator artificial intelligence control system comprises an elevator main control board, a frequency converter, a permanent magnet synchronous motor, an elevator intelligent terminal, an invisible electronic car wall infrared signal generator, an invisible electronic car wall infrared alarm signal receiver, a car speed and posture sensor, a camera, a microphone, an image acquisition and extraction device, a sound acquisition and extraction device, a Hall elevator position sensor, an LED lamp strip and an LED lamp strip control device, and further comprises a remote server.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of platform elevator artificial intelligence control, and provides a platform elevator artificial intelligence control system and method. BACKGROUND

[0002] In recent years, with the improvement of people's living quality, platform elevators are more and more widely used. Compared with traditional elevators, platform elevators have the advantages of small space occupation, low civil engineering requirement, flexible installation and arrangement, etc., and are rapidly developed in the field of home elevators. However, since the platform elevator has no closed car, if the platform elevator is too close to the platform elevator shaft during movement, the contact and extrusion of objects or human bodies with the shaft wall may occur, which may cause danger. Since the platform elevator is a cantilever eccentric load structure, the traditional control method is difficult to guarantee the safety, posture stability and comfort of the elevator operation. SUMMARY

[0003] To solve the above technical problems, the present application provides a platform elevator artificial intelligence control system and artificial intelligence control method to improve the safety, posture stability and comfort of the elevator operation.

[0004] A platform elevator artificial intelligence control system, comprising: an elevator main control board, a frequency converter, a permanent magnet synchronous motor, an elevator intelligent terminal, an invisible electronic car wall infrared signal generator, an invisible electronic car wall infrared alarm signal receiver, a car speed and posture sensor, a camera, a microphone, an image acquisition and extraction device, a sound acquisition and extraction device, a Hall elevator position sensor, an LED lamp strip and an LED lamp strip control device; further comprising a remote server.

[0005] The car operation drive signal output end of the elevator master control board is connected with the car operation drive signal input end of the frequency conversion driver, and the car operation drive signal output end of the frequency conversion driver is connected with the drive signal input end of the permanent magnet synchronous motor; the SPI bus of the frequency conversion driver is connected with the electrical parameter signal output end of the permanent magnet synchronous motor, wherein: the current transformer acquisition circuit for acquiring current is connected in series on the power line of the permanent magnet synchronous motor, and the current acquisition signal output end of the current transformer acquisition circuit is connected with the current acquisition signal input pin of the SPI bus of the frequency conversion driver; the voltage transformer acquisition circuit for acquiring voltage is connected in parallel on the power line of the permanent magnet synchronous motor, and the voltage acquisition signal output end of the voltage transformer acquisition circuit is connected with the voltage acquisition signal input pin of the SPI bus of the frequency conversion driver; the rotary encoder for acquiring the initial position data and real-time position data, i.e. electrical angle signal, of the rotor is arranged on the rotor of the permanent magnet synchronous motor, and the electrical angle signal output end of the rotary encoder is connected with the electrical angle signal input end of the SPI bus of the frequency conversion driver; the electrical parameter signal output port of the frequency conversion driver is connected with the electrical parameter signal input port of the elevator master control board; the elevator master control board is used for calculating the direct-axis and quadrature-axis components according to the input current acquisition signal, voltage acquisition signal and electrical angle signal;

[0006] The signal input end of the invisible electronic car wall infrared signal generator is connected with the invisible electronic car wall signal output end of the elevator master control board; the signal output end of the invisible electronic car wall infrared alarm signal receiver is connected with the invisible electronic car wall alarm signal input end of the elevator master control board; and the electronic car wall alarm signal output end of the elevator master control board is used for connecting the external alarm loudspeaker.

[0007] The signal output end of the car speed and posture sensor is connected with the car speed and posture sensor signal input end of the elevator intelligent terminal; the Ethernet connection port of the elevator intelligent terminal is connected with the elevator server through Ethernet, and is used for uploading the car speed and posture sensor signal.

[0008] The signal output end of the camera is connected with the signal input end of the image acquisition and extraction device, and the signal output end of the image acquisition and extraction device is connected with the image acquisition and face recognition signal input end of the elevator intelligent terminal; the signal output end of the microphone is connected with the signal input end of the sound acquisition and extraction device, and the signal output end of the sound acquisition and extraction device is connected with the sound acquisition and voice recognition signal input end of the elevator intelligent terminal; the control signal input end of the LED lamp strip control device is connected with the LED lamp strip control signal output end of the elevator intelligent terminal; the CAN_BUS interface of the elevator intelligent terminal is connected with the CAN_BUS interface of the elevator master control board through the CAN_BUS bus; and the wireless network interface of the elevator master control board is connected with the wireless network interface of the remote server through the wireless network.

[0009] The signal output end of the Hall elevator position sensor is connected to the elevator position and distance feedback signal input end of the elevator main control board.

[0010] The platform elevator artificial intelligence control system, wherein, comprises:

[0011] The elevator main control board adopts a model XS6; the frequency converter driver adopts a model XT-6; the elevator intelligent terminal adopts a model XEDPU-20; the car speed and posture sensor adopts a model FXAS; the Hall elevator position sensor adopts a model MT6815CT; the current transformer in the current transformer acquisition circuit adopts a model LA-50p; the voltage transformer in the voltage transformer acquisition circuit adopts a model CHV-25P; and the rotary encoder assembled on the rotor of the permanent magnet synchronous motor adopts a model En1387.

[0012] The artificial intelligence control method based on the platform elevator artificial intelligence control system, wherein, comprises the following steps:

[0013] 1) After the system is powered on, self-checking is performed, if there is no exception in the self-checking, the system enters a running state: meanwhile, steps 2), 3), 4), 5) and 6) are entered, if there is an exception in the self-checking, the elevator intelligent terminal displays an alarm source problem, the elevator intelligent terminal uploads alarm information to the elevator main control board, and the elevator main control board uploads the alarm information to a remote server;

[0014] 2) The elevator main control board obtains the absolute position D1 of the elevator in the shaft measured by the Hall elevator position sensor in real time from the elevator position and distance feedback signal input end thereof, the elevator first learns the whole process of the elevator shaft through the Hall elevator position sensor, and the height of each floor is calibrated; the nominal speed D21 of the elevator refers to the highest speed of the elevator in running, the unit is m / s, |D21|≤1.5 m / s, according to the floor height of the elevator, when the distance between the floors where the elevator runs is less than 4 meters, the elevator runs at a low speed D22, the unit of D22 is m / s, |D22|≤0.5 m / s; the unit of the speed acceleration D31 of the elevator in running at the nominal speed is m / s 2 , |D31|≤0.5 m / s 2 ; according to the floor height of the elevator, when the distance between the floors where the elevator runs is less than 4 meters, the low speed acceleration of the elevator in running at the low speed D22 is |D32|≤0.3 / s 2; before each start of the elevator, the destination floor is determined, and according to the comparison result of the destination floor distance and 4 meters, it can be determined whether the speed to be started is the nominal speed D21 or the low speed D22, and whether the acceleration is the nominal acceleration D31 or the low speed acceleration D32; since the elevator is fixed by the brake before starting, it is necessary to have a driving control output to control the elevator to be in a zero-speed stationary state at the moment of opening the brake at the start, but at this time, the elevator is in a stationary state, and various sensors in step 3 cannot collect effective data, and cannot generate driving control output, at this time, a set of initial control parameters is obtained through the driving control model formula (5) to prevent the elevator from losing control at the moment of starting; the driving control model formula (5) is as follows:

[0015]

[0016] 3) When the elevator is running, the frequency converter driver obtains the electrical parameters of the driving permanent magnet synchronous motor in real time through the SPI bus: the current data D4 is collected by the current transformer acquisition circuit, the unit of the current data D4 is A, ampere, and D4<20A; the voltage data D5 is collected by the voltage transformer acquisition circuit, the unit of the voltage data D5 is V, volt, and D5<220V; the rotary encoder assembled on the rotor of the permanent magnet synchronous motor collects the initial position and real-time position of the rotor, that is, the electrical angle data D6, D6=ω*dt, dt is the unit time, and ω is the rotor electrical angular velocity. The unit of the electrical angle D6 is degree γ, and γ<20°;

[0017] After the above data is sent to the elevator main control board, the direct-axis current i d and the cross-axis current i The direct-axis voltage V and the cross-axis voltage V

[0018] Then, the model calculation is entered:

[0019] The mathematical model of the permanent magnet synchronous motor (PMSM) in the synchronous rotating coordinate system is:

[0020]

[0021]

[0022] In the formula: i’ d , i’ q are the decoupling components of the direct-axis and cross-axis currents in the synchronous rotating coordinate system; L d , L qLq, Ld are direct and quadrature axis inductance, respectively, obtained from the motor nameplate; ψ is the permanent magnet flux linkage generated by the rotor magnet, obtained from the motor nameplate; R is the stator winding resistance, obtained from the motor nameplate;

[0023] According to the internal model control principle, the stator current i d q is fully decoupled, and a low-pass feedback filter is added to enhance the robustness of the system, so as to establish the following drive model:

[0024]

[0025]

[0026] wherein: I sin q , I sin d are the output components of the drive model quadrature axis and direct axis current, respectively; I' q , I' d are the input components of the drive model quadrature axis and direct axis current, respectively; α is the modulation rate coefficient, the value range is (0, 1); S is the differential operator, the value is: S = cos(πα), α (0, 1);

[0027] From the permanent magnet synchronous motor drive model, it can be seen that the optimization of drive control is the gain optimization of drive control parameters, and accordingly an adaptive drive control model is established:

[0028]

[0029] wherein: Y(K o ) is the adaptive drive control output, which is output to the frequency converter to drive the permanent magnet synchronous motor and control the running speed and posture of the elevator car;

[0030] d(K i ) is the adaptive drive control disturbance signal input, which is calculated by the following step 6);

[0031] K i is the difference between the real-time value and the ideal value of the posture of the elevator car, and the ideal value of the posture of the elevator car is 1, so:

[0032] K i = D7sin(D10)cosρ + D8sin(D10)cosΦ + D9cos(D10)cosθ - 1;

[0033] I sin q is the calculation result of formula (3); I sin d is the calculation result of formula (4); α x ​Adaptive control coefficient, value range (0, 1)

[0034] K o A set of adaptive drive control parameters output, including proportional coefficient P, integral time I, differential time D; proportional coefficient P, value range: 0~100; integral time I, value range: 20~200ms; differential time D, value range: 20~200ms; K o = P * alpha x + I * (1-alpha x ) + D * (1-alpha x );

[0035] 4) Elevator intelligent terminal real-time acquisition of car attitude parameters from car attitude sensor FXAS, CAN_BUS interface of elevator intelligent terminal uploads current car attitude parameters to elevator main control board through CAN_BUS bus in real time, car attitude parameters include: X axis vibration D7, Y axis vibration D8, Z axis vibration D9, inclination D10, the above parameters are collected by car speed attitude sensor FXAS; the unit of X axis vibration D7 is m / s 2 , |D7|<2m / s 2 ; the unit of Y axis vibration D8 is m / s 2 , |D8|<2m / s 2 ; the unit of Z axis vibration D9 is m / s 2 , |D9|<2m / s 2 ; the unit of inclination D10 is θ degrees, θ<10°;

[0036] 5) Image acquisition and extraction device collects human face and human body image, image pixel is converted into a group of particle swarm with weight and parameter vector;

[0037] 6) The sound acquisition and extraction device collects speech waveform and frequency spectrum at a sampling frequency of 10KHz, and synthesizes speech parameter sequence [O1, O2…O n ];

[0038] Step 7), the attitude estimation model, i.e. submodel M1, reads the components A X , A Y , A Z of X, Y, Z axes by car speed attitude sensor FXAS, considering that the elevator car is constrained in the vertical direction by the elevator shaft guide rail, the description of the elevator car attitude corresponds to the relative vertical direction angles of X, Y, Z axes as ρ, Φ, θ;

[0039] Among them:

[0040] The difference between the real-time value and the ideal value of the elevator car attitude is used as the disturbance feedback input d(Ki ), constituting a closed loop feedback; K i = D7sin(D10)cos p + D8sin(D10)cos + D9cos(D10)cos -1;

[0041] D9 cos(D10)cos0-1;

[0042] Face and human behavior feature recognition model, i.e. sub-model M2, converts the face and human image pixels collected by the image collection and extraction device into a group of particle swarm with flight inertia weight and parameter vector. Flight inertia weight w represents the activity degree or the ability to inherit the previous speed, and the value range is [0, 1], the greater the value, the more active. The parameter vector includes ion number, individual historical position and speed value. Each particle has two attributes of position x and speed v. The position x takes the image center as the coordinate origin, and the vertical and horizontal coordinates are labeled with the basic unit of the minimum particle distance. The vertical coordinate value range is [-640, 640], and the horizontal coordinate value range is [-400, 400]. The speed size is the displacement of the particle within the time interval, and the direction is from the starting time position to the ending time position. After the conversion of multiple image pixels, the position and speed of each particle are iteratively updated on the time domain axis. The particles share information to record the parameter information of the whole group. These information includes the position and speed of particles from 1 to n at different historical time intervals, the position and speed of particles from 1 to n at the same time and different dimensions, and the dimension refers to a variable of the search function. By searching the historical optimal value recorded by each particle, the particle records parameters including flight inertia weight w, position x and speed v. These basic data are stored in the database in the form of two-dimensional array with historical time and dimension as two variables, and then iteratively calculated by formula (6) and (7) to finally obtain the optimal solution of particle position, i.e. the minimum position value.

[0043] Particle velocity iteration:

[0044] v k id = w i v k-1 id + c1rand1( p i best-x k-1 id )+c2rand2( p i best-x k-1 id ) (6)

[0045] wherein: v k idA variable of the search function refers to a dimension for a flight speed of the particle i in the d dimension in the k iteration process, d=2; k=50 is the iteration number; i=1~2000 sequentially marks 2000 particles; c1=c2=2 is a learning factor of the particle, the learning factor expresses particle group diversity, and the learning factor takes a value [0, 4], 0 represents loss of group diversity, and 4 represents maximum group diversity; rand1 and rand2 are randomly taken in a range from 0 to 1, and are used to increase randomness of the search; w i =0.6 represents a flight inertia weight, the flight inertia weight expresses a particle activity degree or a capability of inheriting a previous speed, and the flight inertia weight takes a value range [0, 1], and the greater the value is, the more active the particle is; p i best represents a historical optimal solution of the particle i in the d dimension in the k-1 iteration, that is, a smaller one of two adjacent particle position iteration results, and the historical optimal solution is stored in a database in a one-dimensional array form with a historical time as a variable;

[0046] Particle position iteration:

[0047] x k id = x k-1 id +v k id (7)

[0048] Wherein: x k id A position of the particle i in the d dimension in the k iteration process, that is, a distance from an image center as a coordinate origin, is arranged according to distance values in different dimensions, and a minimum value x i of the particle i from the origin is solved, that is, an optimal solution of the i-th particle position, and an optimal solution of each particle in the particle group is solved, and accordingly, higher image recognition precision and error correction capability can be obtained, and a best face and human body image is reconstructed, and face recognition and lift taking behavior judgment are performed;

[0049] A speech perception linear prediction model, that is, a sub-model M3, a sound collection and extraction device processes speech into a speech parameter feature vector frame sequence [O1, O2…O n ], which contains a speech waveform and a spectrum, a speech sampling frequency is 10KHz, a frame length is 20ms, n=200, and after the elevator intelligent terminal receives the speech parameters, a current prediction value is solved according to the following prediction model:

[0050]

[0051] In the formula: a jFor the prediction coefficient, the coefficient is solved by the method of windowing the speech parameter feature vector frame sequence by autocorrelation method, G is a gain factor, the factor is the mean value of the square of the mean square deviation of the speech parameter feature vector, p is the prediction order, the larger the value, the closer the prediction model to the original speech signal, the value is not less than 12, e(n) is the prediction error, the error is the same as the frequency of the original speech signal, and the amplitude is less than 25% of the amplitude of the original speech signal; compare the predicted value with the learning value, and identify the speech information result;

[0052] The elevator safety evaluation model, i.e. the sub-model M4, is to establish an evaluation model for the safety margin and its weight of the parameters affecting the elevator running quality, so as to obtain the system safety margin in combination with the running conditions, design parameters, risk evaluation and other factors of the system:

[0053] △R=W x |f x (i0,j0)-f x (i m ,j m )|+W y |f y (i0,j0)-f y (i m ,j m )|

[0054] +W z |f z (i0,j0)-f z (i m ,j m )| (10)

[0055] In the formula: △R is the safety margin of the elevator system, W x , W y , W z are respectively the safety margin weights of the three-dimensional vector parameters of the elevator, according to the platform structure characteristics, dragging mode, guiding and braking characteristics of the platform elevator, W x =0.5, W y =0.2, W z =0.3, f x (i0,j0), f y (i0,j0), f z (i0,j0) are respectively the national standard values of the three-dimensional vector parameters of the elevator running quality, which are fixed values under different speeds and different loads, f x (i m ,j m ), f y (i m ,j m ), f z (i m ,jm ) are the maximum disturbance values of the three-dimensional vector parameters of the elevator load, which are obtained by taking the maximum value of the real-time measurement of the car speed and attitude sensor FXAS during the operation of the elevator; the three-dimensional vector parameters affecting the elevator load include speed, acceleration, vibration and car inclination;

[0056] Step 8), the adaptive drive control output is obtained from the model formula (5) to obtain a set of adaptive drive control parameters, including the proportional coefficient P, the integral time I and the differential time D;

[0057]

[0058] K o = P * alpha x + I * (1-alpha x ) + D * (1-alpha x );

[0059] When the system is powered on for the first time, a set of initial control parameters are obtained, in which the integral time and the differential time are both fixed median values of 110 ms, and the proportional coefficient is a random value, then the system dynamically adjusts according to the closed-loop feedback results of the elevator load in the order of proportional coefficient, integral time and differential time, and gradually approaches a set of optimal values, according to the characteristics of the platform elevator load inertia, the adjustment ranges of the three parameters are: proportional coefficient: 0~100, integral time: 20~200 ms, differential time: 20~200 ms; the initial values are: proportional coefficient: 50, integral time: 80 ms, differential time: 80 ms; the basis for dynamic adjustment is that under the premise of meeting or being better than the national standard value of the elevator load, the proportional coefficient tends to 100, the integral time tends to 20 ms, and the differential time tends to 20 ms, and the system control is in the optimal state; the image recognition accuracy is greater than 95%, and the voice recognition accuracy is greater than 98%;

[0060] The elevator main control board transmits the proportional coefficient P, the integral time I and the differential time D drive control parameters to the frequency converter, drives the permanent magnet synchronous motor, and drags the elevator to run, at the same time, every 200 ms, these drive control parameters, elevator state parameters, image and voice interaction information, fault alarm code are uploaded to the remote server to establish the elevator operation database.

[0061] The platform elevator artificial intelligence control system and the artificial intelligence control method based on the platform elevator artificial intelligence control system provided by the application obtain various required data through the invisible electronic car wall infrared signal generator, the invisible electronic car wall infrared alarm signal receiver, the car speed and attitude sensor, the camera, the microphone, the image acquisition and extraction device, the sound acquisition and extraction device and the Hall elevator position sensor, and after the data are sent to the elevator main control board for calculation and processing, the elevator feedback control is performed, and the safety, attitude stability and comfort of the elevator operation are improved. BRIEF DESCRIPTION OF DRAWINGS

[0062] Figure 1 The structural principle diagram of the platform elevator artificial intelligence control system is shown in the figure.

[0063] Figure 2 The schematic diagram of the relative vertical direction angle p, f, q of the elevator car posture corresponding to the X, Y, Z axis is shown in the figure. DETAILED DESCRIPTION

[0064] The present application provides a platform elevator artificial intelligence control system, as shown in the figure, comprising: a platform elevator artificial intelligence control system, comprising: an elevator main control board, a frequency converter driver, a permanent magnet synchronous motor, an elevator intelligent terminal, an invisible electronic car wall infrared signal generator, an invisible electronic car wall infrared alarm signal receiver, a car speed posture sensor, a camera, a microphone, an image acquisition and extraction device, a sound acquisition and extraction device, a Hall elevator position sensor, an LED lamp strip and an LED lamp strip control device; further comprising a remote server. Figure 1

[0065] The car running drive signal output end of the elevator main control board is connected to the car running drive signal input end of the frequency converter driver, and the car running drive signal output end of the frequency converter driver is connected to the drive signal input end of the permanent magnet synchronous motor; the SPI bus of the frequency converter driver is connected to the electrical parameter signal output end of the permanent magnet synchronous motor, wherein: a current transformer acquisition circuit for acquiring current is connected in series on the power supply line of the permanent magnet synchronous motor, and the current acquisition signal output end of the current transformer acquisition circuit is connected to the current acquisition signal input pin of the SPI bus of the frequency converter driver; a voltage transformer acquisition circuit for acquiring voltage is connected in parallel on the power supply line of the permanent magnet synchronous motor, and the voltage acquisition signal output end of the voltage transformer acquisition circuit is connected to the voltage acquisition signal input pin of the SPI bus of the frequency converter driver; a rotary encoder for acquiring rotor initial position data and real-time position data, i.e. electrical angle signal, is arranged on the rotor of the permanent magnet synchronous motor, and the electrical angle signal output end of the rotary encoder is connected to the rotary encoder electrical angle signal input end of the SPI bus of the frequency converter driver; the electrical parameter signal output port of the frequency converter driver is connected to the frequency converter electrical parameter signal input port of the elevator main control board; the elevator main control board is used for calculating the direct-axis and quadrature-axis components according to the input current acquisition signal, voltage acquisition signal and electrical angle signal.

[0066] The signal input end of the invisible electronic car wall infrared signal generator is connected to the invisible electronic car wall signal output end of the elevator main control board; the signal output end of the invisible electronic car wall infrared alarm signal receiver is connected to the invisible electronic car wall alarm signal input end of the elevator main control board; and the electronic car wall alarm signal output end of the elevator main control board is used for connecting an external alarm loudspeaker. ​

[0067] The signal output end of the car speed posture sensor is connected to the car speed posture sensor signal input end of the elevator intelligent terminal; the Ethernet connection port of the elevator intelligent terminal is connected to the elevator server through Ethernet, for uploading the car speed posture sensor signal;

[0068] The signal output end of the camera is connected to the signal input end of the image acquisition and extraction device, and the signal output end of the image acquisition and extraction device is connected to the image acquisition and face recognition signal input end of the elevator intelligent terminal; the signal output end of the microphone is connected to the signal input end of the sound acquisition and extraction device, and the signal output end of the sound acquisition and extraction device is connected to the sound acquisition and voice recognition signal input end of the elevator intelligent terminal; the control signal input end of the LED lamp strip control device is connected to the LED lamp strip control signal output end of the elevator intelligent terminal; the CAN_BUS interface of the elevator intelligent terminal is connected to the CAN_BUS interface of the elevator main control board through the CAN_BUS bus; the wireless network interface of the elevator main control board is connected to the wireless network interface of the remote server through the wireless network;

[0069] The signal output end of the Hall elevator position sensor is connected to the elevator position and distance feedback signal input end of the elevator main control board.

[0070] The platform elevator artificial intelligence control system, wherein, comprising:

[0071] The elevator main control board adopts a model of XS6; the frequency converter driver adopts a model of XT-6; the elevator intelligent terminal adopts a model of XEDPU-20; the car speed posture sensor adopts a model of FXAS; the Hall elevator position sensor adopts a model of MT6815CT; the current transformer in the current transformer acquisition circuit adopts a model of LA-50p; the voltage transformer in the voltage transformer acquisition circuit adopts a model of CHV-25P; and the rotary encoder assembled on the rotor of the permanent magnet synchronous motor adopts a model of En1387.

[0072] The artificial intelligence control method based on the platform elevator artificial intelligence control system, wherein, comprising the following steps:

[0073] 1) After the system is powered on, self-checking, if there is no exception in self-checking, entering the running state: at the same time, entering steps 2), 3), 4), 5), 6), if there is an exception in self-checking, the elevator intelligent terminal displays the alarm source problem, the elevator intelligent terminal uploads the alarm information to the elevator main control board, and the elevator main control board uploads the alarm information to the remote server;

[0074] 2) Elevator main control board from its elevator position and distance feedback signal input real-time acquisition of Hall elevator position sensor measured in the elevator shaft absolute position D1, the elevator first through the Hall elevator position sensor on the elevator shaft throughout the learning, each floor height calibration; elevator nominal speed D21 refers to the highest speed of the elevator, units of m / s, |D21|≤1.5 m / s, according to the elevator floor height, when the distance between the elevator running floors less than 4 meters, the elevator runs at low speed D22, D22 units of m / s, |D22|≤0.5 m / s; the speed acceleration D31 units of m / s 2 , |D31|≤0.5 m / s 2 ; according to the elevator floor height, when the distance between the elevator running floors less than 4 meters, the elevator runs at low speed D22, the low speed acceleration is |D32|≤0.3 / s 2 ; before each start of the elevator, the destination floor is determined, according to the destination floor distance and 4 meters comparison results, can determine the speed of the start of the running is the nominal speed D21 or low speed D22, acceleration is the speed acceleration D31 or low speed acceleration D32; due to the elevator before starting by the brake fixed, starting brake open moment must have drive control output control elevator in zero speed static state, but, at this time the elevator is in a static state, step 3 of various sensors can not collect valid data, can not produce drive control output, at this time through the drive control model formula (5) to get a set of initial control parameters, to prevent the elevator starting moment out of control; the drive control model formula (5) is as follows:

[0075]

[0076] 3) When the elevator is running, the frequency converter driver real-time acquires the electrical parameters of the driving permanent magnet synchronous motor through the SPI bus: the current transformer acquisition circuit acquires current data D4, the unit of current data D4 is A, ampere, D4<20A; the voltage transformer acquisition circuit acquires voltage data D5, the unit of voltage data D5 is V, volt, D5<220V; the rotary encoder assembled on the rotor of the permanent magnet synchronous motor acquires the initial position and real-time position of the rotor, i.e. electrical angle data D6, D6=ω*dt, dt is the unit time, ω is the rotor electrical angular velocity, the unit of electrical angle D6 is degree γ, γ<20°;

[0077] After the above data is sent to the elevator main control board, the direct-axis current i d =D4*cos(45°-D6)*cos(D6), the cross-axis current The direct-axis voltage is calculated by the voltage data D5 and the electrical angle data D6:

[0078] After that, into the model calculation:

[0079] Permanent magnet synchronous motor (PMSM) in the synchronous rotating coordinate system mathematical model is:

[0080]

[0081]

[0082] In the formula: i' d , i' q The decoupling components of the direct axis and the cross axis current in the synchronous rotating coordinate system; L d , L q The direct axis and cross axis inductance, obtained from the motor nameplate; ψ is the permanent magnet flux linkage generated by the rotor magnetic steel, obtained from the motor nameplate; R is the stator winding resistance, obtained from the motor nameplate;

[0083] According to the internal model control principle, the stator current i d , i q Completely decoupled, add a low-pass feedback filter to enhance the robustness of the system, that is, the following drive model can be established:

[0084]

[0085]

[0086] In the formula: I sin q , I sin d The output components of the drive model cross axis and direct axis current; I' q , I' d The input components of the drive model cross axis and direct axis current; α is the modulation rate coefficient, the value range (0, 1); S is the differential operator, the value: S = cos(πα), α (0, 1);

[0087] From the permanent magnet synchronous motor drive model, the optimization of drive control is the gain optimization of drive control parameters, and accordingly an adaptive drive control model is established:

[0088]

[0089] In the formula: Y(K o ) is the adaptive drive control output, output to the frequency converter to drive the permanent magnet synchronous motor, control the elevator car running speed and running posture;

[0090] d(K i) is the adaptive drive control disturbance signal input, which is calculated by step 6) below;

[0091] K i is the real-time value of the elevator car attitude, and is the difference between the real-time value and the ideal value of the elevator car attitude, the ideal value of the elevator car attitude is 1, then:

[0092] K i = D7sin(D10)cosθ + D8sin(D10)cosΦ + D9cos(D10)cosθ - 1;

[0093] I sin q is the calculation result of formula (3); I sin d is the calculation result of formula (4); α x is an adaptive control coefficient, the value range is (0, 1);

[0094] K o is a set of adaptive drive control parameters output, including a proportional coefficient P, an integral time I, and a differential time D; the proportional coefficient P, the value range is: 0~100; the integral time I, the value range is: 20~200ms; the differential time D, the value range is: 20~200ms; K o = P*α x + I*(1-α x )+ D*(1-α x );

[0095] 4) The elevator intelligent terminal obtains the car attitude parameters transmitted by the car attitude sensor FXAS in real time, the CAN_BUS interface of the elevator intelligent terminal uploads the current car attitude parameters to the elevator main control board in real time through the CAN_BUS bus, and the car attitude parameters include: X-axis vibration D7, Y-axis vibration D8, Z-axis vibration D9, and inclination D10, the above parameters are collected by the car attitude sensor FXAS; the unit of X-axis vibration D7 is m / s 2 , |D7|<2m / s 2 ; the unit of Y-axis vibration D8 is m / s 2 , |D8|<2m / s 2 ; the unit of Z-axis vibration D9 is m / s 2 , |D9|<2m / s 2 ; the unit of inclination D10 is θ degrees, θ<10°;

[0096] 5) The image acquisition and extraction device acquires the face and body images of a person, and converts the image pixels into a set of particle groups with weight and parameter vectors;

[0097] 6), the sound acquisition and extraction device acquires speech waveform and spectrum at a sampling frequency of 10KHz, and synthesizes speech parameter sequence [O1, O2...O n ];

[0098] Step 7), the attitude estimation model, i.e. the sub-model M1, reads the components A X , A Y , A Z of the space X, Y, Z axes from the car speed attitude sensor FXAS, and describes the attitude of the elevator car, i.e. the relative vertical direction angles of the X, Y, Z axes, as ρ, Φ, θ;

[0099] Wherein:

[0100] The difference between the real-time value and the ideal value of the attitude of the elevator car is taken as the disturbance feedback input d(K i ) of the elevator drive control, and constitutes a closed-loop feedback; K i =D7sin(D10)cosρ+D8sin(D10)cosΦ+D9cos(D10)cosθ-1;

[0101] Figure 2 A schematic diagram for describing the attitude of the elevator car, i.e. the relative vertical direction angles of the X, Y, Z axes, ρ, Φ, θ;

[0102] The face and human behavior feature recognition model, i.e. the sub-model M2, converts the face and human image pixels collected by the image collection and extraction device into a group of particle swarm with flight inertia weight and parameter vector. The flight inertia weight w represents the activity degree or the ability to inherit the previous speed of the particle, and the value range is [0, 1], the greater the value, the more active. The parameter vector includes ion number, individual historical position and speed value. Each particle has two attributes of position x and speed v. The position x takes the image center as the coordinate origin, and the vertical and horizontal coordinates are labeled with the basic unit of the minimum particle distance. The vertical coordinate value range is [-640, 640], and the horizontal coordinate value range is [-400, 400]. The speed size is the displacement of the particle within the time interval, and the direction is from the starting time position to the ending time position. The multi-frame image pixels are iteratively updated in the time domain axis, and the particles share information to record the parameter information of the whole group. These information includes the position and speed of particles from 1 to n at different historical time intervals, the position and speed of particles from 1 to n at the same time, and the dimension of the search function. By searching the respective historical optimal value recorded by each particle, the particle records the parameters including flight inertia weight w, position x and speed v. These basic data are stored in the database in the form of two-dimensional array with historical time and dimension as two variables, and then iteratively calculated by formula (6) and (7) to finally obtain the optimal solution of particle position, i.e. the minimum position value.

[0103] Particle speed iteration:

[0104] v k id = w i v k-1 id + c1rand1( p i best-x k-1 id )+c2rand2( p i best-x k-1 id ) (6)

[0105] wherein: v k id is the flight speed of particle i in d dimension in the kth iteration process, a variable of the search function refers to a dimension, d = 2; k = 50 is the iteration number; i = 1 ~ 2000 marks 2000 particles in order; c1 = c2 = 2 is the learning factor of the particle, which represents the diversity of the particle group, and the value is [0, 4], 0 represents the loss of group diversity, and 4 represents the maximum group diversity; rand1 and rand2 are randomly taken from 0 to 1, which is used to increase the randomness of the search; w i=0.6 represents the flight inertia weight, the flight inertia weight represents the activity degree or the ability of inheriting the previous speed, the value range is [0, 1], the greater the value, the more active; p i Best represents the historical optimal solution of the particle i in the d-dimensional dimension in k-1 iterations, that is, the smaller one of the adjacent two particle position iteration results, which is stored in the database in the form of a one-dimensional array with the historical time as the variable;

[0106] Particle position iteration:

[0107] x k id = x k-1 id + v k id (7)

[0108] Wherein: x k id The position of particle i in d-dimensional in the kth iteration process, that is, the distance from the image center as the coordinate origin, the distance value under different dimensions is arranged, and the minimum value x i , that is, the optimal solution of the i-th particle position, the optimal solution of each particle in the particle group is solved, and higher image recognition accuracy and error correction ability can be obtained, and the best face and human body image is reconstructed, and face recognition and lift behavior judgment is performed;

[0109] The speech perception linear prediction model is a sub-model M3, and the sound collection and extraction device processes the voice into a voice parameter feature vector frame sequence [O1, O2…O n ], which contains voice waveform and spectrum, the voice sampling frequency is 10KHz, the frame length is 20ms, then n=200, and the elevator intelligent terminal receives the voice parameters and calculates the current prediction value according to the following prediction model:

[0110]

[0111] In the formula: a j is a prediction coefficient, which is solved by the method of windowing the voice parameter feature vector frame sequence by autocorrelation, G is a gain factor, which is the average value of the square of the mean square deviation of the voice parameter feature vector, p is the prediction order, the greater the value, the closer the prediction model to the original voice signal, the value is not less than 12, e(n) is the prediction error, which has the same frequency as the original voice signal, and the amplitude is less than 25% of the amplitude of the original voice signal; compare the prediction value with the learning value, and identify the voice information result;

[0112] The elevator safety evaluation model, i.e., the sub-model M4, is to establish an evaluation model for the safety margin of the parameters affecting the elevator ride quality and the weight thereof, to obtain the system safety margin in combination with the operation conditions, design parameters, risk evaluation and other factors of the system:

[0113] △R=W x |f x (i0,j0)-f x (i m ,j m )|+W y |f y (i0,j0)-f y (i m ,j m )|

[0114] +W z |f z (i0,j0)-f z (i m ,j m )| (10)

[0115] In the formula, △R is the safety margin of the elevator system, W x , W y , and W z are the safety margin weights of the three-dimensional vector parameters of the elevator, respectively, W x =0.5, W y =0.2, and W z =0.3 are taken according to the platform structure characteristics, dragging mode, guiding and braking characteristics of the platform elevator, f x (i0,j0), f y (i0,j0), and f z (i0,j0) are the standard values of the three-dimensional vector parameters of the elevator ride quality according to the national standard, which are fixed values under different speeds and different loads, f x (i m ,j m ), f y (i m ,j m ), and f z (i m ,j m ) are the maximum disturbance values of the three-dimensional vector parameters of the elevator ride quality, which are obtained by measuring the values in real time through the car speed attitude sensor FXAS during the elevator operation and then finding the maximum value; the three-dimensional vector parameters affecting the elevator ride quality include speed, acceleration, vibration and car inclination;

[0116] Step 8), the adaptive drive control output is obtained from the model formula (5) to obtain a set of adaptive drive control parameters, including the proportional coefficient P, the integral time I, and the differential time D;

[0117]

[0118] K o = P*alpha x + I*(1-alpha x )+ D*(1-alpha x );

[0119] The system first gets initial control parameters with integral time, differential time being fixed median 110ms and proportional coefficient being random value when the system is powered on for the first time, then the system adjusts dynamically according to the closed-loop feedback result of the elevator load according to the order of proportional coefficient, integral time and differential time, and gradually approaches a set of optimal values according to the characteristics of the platform elevator load inertia, the adjustment ranges of the three parameters are: proportional coefficient: 0-100, integral time: 20-200ms, differential time: 20-200ms; the initial values are: proportional coefficient: 50, integral time: 80ms, differential time: 80ms; the basis of dynamic adjustment is that the proportional coefficient approaches 100, the integral time approaches 20ms and the differential time approaches 20ms under the premise of meeting or being better than the national standard value of the elevator load, and the system control is in the optimal state; the image recognition accuracy is greater than 95% and the voice recognition accuracy is greater than 98%;

[0120] The elevator main control board transmits the proportional coefficient P, integral time I and differential time D driving control parameters to the frequency converter, drives the permanent magnet synchronous motor and drags the elevator to run, at the same time, the driving control parameters, elevator state parameters, image and voice interaction information and fault alarm code are uploaded to the remote server every 200ms to establish the elevator operation database.

[0121] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited to this, any skilled person in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application; therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A platform elevator artificial intelligence control system, characterized by, The elevator main control board, the frequency conversion driver, the permanent magnet synchronous motor, the elevator intelligent terminal, the invisible electronic car wall infrared signal generator, the invisible electronic car wall infrared alarm signal receiver, the car speed and posture sensor, the camera, the microphone, the image acquisition and extraction device, the sound acquisition and extraction device, the Hall elevator position sensor, the LED lamp strip and the LED lamp strip control device are connected; the remote server is also included; The car running drive signal output end of the elevator main control board is connected with the car running drive signal input end of the frequency conversion driver, and the car running drive signal output end of the frequency conversion driver is connected with the drive signal input end of the permanent magnet synchronous motor; The SPI bus of the frequency conversion driver is connected with the electrical parameter signal output end of the permanent magnet synchronous motor, wherein: the current transformer acquisition circuit for acquiring current is connected in series on the power line of the permanent magnet synchronous motor, and the current acquisition signal output end of the current transformer acquisition circuit is connected with the current acquisition signal input pin of the SPI bus of the frequency conversion driver; the voltage transformer acquisition circuit for acquiring voltage is connected in parallel on the power line of the permanent magnet synchronous motor, and the voltage acquisition signal output end of the voltage transformer acquisition circuit is connected with the voltage acquisition signal input pin of the SPI bus of the frequency conversion driver; the rotating encoder for acquiring the initial position data and real-time position data, i.e. the electrical angle signal, of the rotor is arranged on the rotor of the permanent magnet synchronous motor, and the electrical angle signal output end of the rotating encoder is connected with the electrical angle signal input end of the rotating encoder of the SPI bus of the frequency conversion driver; the electrical parameter signal output port of the frequency conversion driver is connected with the frequency conversion driver electrical parameter signal input port of the elevator main control board; the elevator main control board is used for calculating the direct-axis and quadrature-axis components according to the input current acquisition signal, voltage acquisition signal and electrical angle signal; The signal input end of the invisible electronic car wall infrared signal generator is connected with the invisible electronic car wall signal output end of the elevator main control board; the signal output end of the invisible electronic car wall infrared alarm signal receiver is connected with the invisible electronic car wall alarm signal input end of the elevator main control board; and the electronic car wall alarm signal output end of the elevator main control board is used for connecting the external alarm loudspeaker; The signal output end of the car speed and posture sensor is connected with the car speed and posture sensor signal input end of the elevator intelligent terminal; the Ethernet connection port of the elevator intelligent terminal is connected with the elevator server through Ethernet, and is used for uploading the car speed and posture sensor signal; The signal output end of the camera is connected with the signal input end of the image acquisition and extraction device, the signal output end of the image acquisition and extraction device is connected with the image acquisition and face recognition signal input end of the elevator intelligent terminal; the signal output end of the microphone is connected with the signal input end of the sound acquisition and extraction device, the signal output end of the sound acquisition and extraction device is connected with the sound acquisition and voice recognition signal input end of the elevator intelligent terminal; the control signal input end of the LED lamp strip control device is connected with the LED lamp strip control signal output end of the elevator intelligent terminal; the CAN_BUS interface of the elevator intelligent terminal is connected with the CAN_BUS interface of the elevator main control board through the CAN_BUS bus; and the wireless network interface of the elevator main control board is connected with the wireless network interface of the remote server through the wireless network. ​ The signal output end of the Hall elevator position sensor is connected to the elevator position and distance feedback signal input end of the elevator main control panel.

2. The platform elevator artificial intelligence control system of claim 1, wherein, It comprises: The elevator main control panel adopts the model XS6; the frequency conversion driver adopts the model XT-6; The elevator intelligent terminal adopts the model XEDPU-20; the car speed and posture sensor adopts the model FXAS; the Hall elevator position sensor adopts the model MT6815CT; the current transformer in the current transformer acquisition circuit adopts the model LA-50p; the voltage transformer in the voltage transformer acquisition circuit adopts the model CHV-25P; and the rotary encoder assembled on the rotor of the permanent magnet synchronous motor adopts the model En1387.

3. An artificial intelligence control method based on the platform elevator artificial intelligence control system of claim 1 or 2, characterized by, It comprises the following steps: 1) After the system is powered on, self-checking is performed, and if there is no abnormality, the system enters a running state; meanwhile, steps 2), 3), 4), 5) and 6) are entered; if there is an abnormality, the elevator intelligent terminal displays an alarm source problem, the elevator intelligent terminal uploads alarm information to the elevator main control panel, and the elevator main control panel uploads the alarm information to a remote server; 2) The elevator main control panel obtains the absolute position D1 of the elevator in the shaft measured by the Hall elevator position sensor in real time from the elevator position and distance feedback signal input end thereof, and the elevator learns the whole process of the elevator shaft through the Hall elevator position sensor and calibrates the height of each floor; The nominal speed D21 of the elevator refers to the highest speed of the elevator, and the unit is m / s, |D21|≤1.5 m / s; according to the floor height of the elevator, when the distance between the floors on which the elevator runs is less than 4 meters, the elevator runs at a low speed D22, and the unit of D22 is m / s, |D22|≤0.5 m / s; The unit of the acceleration D31 at the rated speed of the elevator is m / s 2 , |D31|≤0.5m / s 2 ; according to the floor height of the elevator, when the distance between the floors where the elevator runs is less than 4 meters, the acceleration D32 of the elevator running at the low speed D22 is |D32|≤0.3 / s 2 ; Before the elevator starts each time, the destination floor is determined, and according to the comparison result of the distance of the destination floor and 4 meters, it can be determined whether the speed to be started to run is the nominal speed D21 or the low speed D22, and whether the acceleration is the low speed acceleration D32 or the low speed acceleration D32; since the elevator is fixed by the brake before starting, it is necessary to have a drive control output to control the elevator to be in a zero-speed stationary state at the moment when the brake is opened, but at this time, the elevator is in a stationary state, and various sensors in step 3 cannot collect effective data, and cannot generate a drive control output, so an initial control parameter is obtained through a drive control model to prevent the elevator from losing control at the moment of starting; 3) When the elevator is running, the frequency converter obtains the electrical parameters of the driving permanent magnet synchronous motor in real time through the SPI bus: the current transformer acquisition circuit acquires current data D4, the unit of the current data D4 is A, ampere, D4 <20A; the voltage transformer acquisition circuit acquires voltage data D5, the unit of the voltage data D5 is V, volt, D5 <220V; the rotary encoder assembled on the rotor of the permanent magnet synchronous motor acquires the initial position and real-time position of the rotor, i.e. electrical angle data D6, , is the unit time, is the rotor electric angular velocity, the unit of the electrical angle D6 is degree γ, γ <20 o ; After the above data is fed into the elevator main control board, the direct-axis current , the quadrature-axis current are calculated from the current data D4 and the electrical angle data D6; the direct-axis voltage , the quadrature-axis voltage are calculated from the voltage data D5 and the electrical angle data D6; Then, model calculation is performed: The mathematical model of the permanent magnet synchronous motor (PMSM) in the synchronous rotating coordinate system is: (1) (2) wherein: are the decoupled components of the direct and quadrature axis currents in the synchronous rotating reference frame, respectively; are the direct and quadrature axis inductances, obtained from the motor nameplate, respectively; is the permanent magnet flux linkage due to the rotor magnets, obtained from the motor nameplate; is the stator winding resistance, obtained from the motor nameplate; According to the internal model control principle, the direct-axis current , the quadrature-axis current is completely decoupled, and a low-pass feedback filter is added to enhance the robustness of the system. Thus, the following drive model is established: (3) (4) wherein: are the output components of the drive model quadrature and direct axis currents, respectively; are the input components of the drive model quadrature and direct axis currents, respectively; is the modulation rate coefficient, taking values in the range (0, 1); is the differential operator, taking values: , (0, 1); As can be seen from the permanent magnet synchronous motor drive model, the optimization of the drive control is the gain optimization of the drive control parameter, and accordingly, an adaptive drive control model is established: (5) In the formula: is an adaptive drive control output, output to a variable frequency drive to drive a permanent magnet synchronous motor, control the elevator car running speed and running posture; an adaptive drive control disturbance signal input; is the difference between the real value of the elevator car attitude and the ideal value of the elevator car attitude, the ideal value of the elevator car attitude is 1, then: ； is the calculation result of formula (3); is the calculation result of formula (4); is an adaptive control coefficient, with a value range of (0, 1); A set of adaptive drive control parameters, including proportional coefficient P, integral time I, and differential time D, are output. The proportional coefficient P has a value range of 0-100; the integral time I has a value range of 20-200 ms; and the differential time D has a value range of 20-200 ms. ; 4) The elevator intelligent terminal obtains the car attitude parameters from the car speed attitude sensor FXAS in real time, the CAN_BUS interface of the elevator intelligent terminal uploads the current car attitude parameters to the elevator main control board through the CAN_BUS bus in real time, and the car attitude parameters include: X-axis vibration D7, Y-axis vibration D8, Z-axis vibration D9, and inclination D10, wherein the above parameters are collected by the car speed attitude sensor FXAS; the unit of X-axis vibration D7 is m / s 2 , |D7| < 2 m / s 2 ; the unit of Y-axis vibration D8 is m / s 2 , |D8| < 2 m / s 2 ; the unit of Z-axis vibration D9 is m / s 2 , |D9| < 2 m / s 2 ; the unit of inclination D10 is θ degrees, and θ < 10 o ; 5) The image acquisition and extraction device acquires the face and body images of a person, and converts the image pixels into a group of particle swarms with weight and parameter vectors; 6) The sound acquisition and extraction device acquires the speech waveform and spectrum at a sampling frequency of 10 KHz, and synthesizes the speech parameter sequence ; Step 7), the attitude estimation model, i.e., the sub-model M1, reads the components A of the spatial X, Y, Z axes from the car speed attitude sensor FXAS X Y Z , taking into account that the elevator car is constrained in the vertical direction by the elevator shaft guide rails, the relative vertical angles of the elevator car attitude, i.e., the angles of the X, Y, Z axes, are described as ;​​​​ wherein: , a difference between a real-time value and an ideal value of an elevator car attitude as a disturbance feedback input for elevator drive control , constituting a closed-loop feedback; ; The face and human behavior feature recognition model, i.e. the sub-model M2, converts the face and human image pixels collected by the image collection and extraction device into a group of particle swarm with flight inertia weight and parameter vector. The flight inertia weight represents the expression of particle activity or the ability to inherit previous speed, and the value range is [0, 1]. The greater the value, the more active the particle is. The parameter vector includes ion number, individual historical position and speed value. Each particle has position and speed two attributes, position is taken as the coordinate origin, and the vertical and horizontal coordinates are marked with the minimum particle distance as the basic unit. The vertical coordinate value range is [-640, 640], and the horizontal coordinate value range is [-400, 400]. The speed size is the displacement of the particle within the time interval, and the direction is the position from the starting time to the ending time. After the pixel conversion of multiple images, the position and speed of the particle are iteratively updated on the time domain axis. The particles share information to record the parameter information of the whole group. These information includes the position and speed of the particles from 1 to n at different historical time intervals, the position and speed of the particles from 1 to n at the same time in different dimensions, and the dimension refers to a variable of the search function. By searching the respective historical optimal value recorded by each particle, the parameters recorded by the particle include flight inertia weight , position and speed . These basic data are stored in the database in the form of two-dimensional array with two variables of historical time and dimension, and then iteratively calculated by formula to finally obtain the optimal solution of the particle position, i.e. the minimum position value. Particle velocity iteration: ； Wherein: is the particle in the th iteration process dimensional flight speed, one variable of the search function refers to a dimension, = 2; = 50 is the number of iterations; = 1~2000 is the order mark of 2000 particles; is the learning factor of the particle, which expresses the particle swarm diversity, and the value is [0, 4], 0 represents the loss of group diversity, and 4 represents the maximum group diversity; is randomly taken from the range between 0 and 1, which is used to increase the randomness of the search; = 0.6 represents the flight inertia weight, which expresses the activity degree or the ability to inherit the previous speed of the particle, and the value range is [0, 1], the larger the value, the more active; represents the particle is th iteration in the dimensional historical optimal solution, that is, the smaller one of the iteration results of the positions of the particles adjacent to each other, which is stored in the database in the form of one-dimensional array with the historical time as a variable; Particle position iteration: ； Wherein: The first Particle in the secondary iteration process In The position of the dimension, that is, the distance from the image center as the coordinate origin, is arranged according to the distance value under different dimensions, and the position of the particle The minimum value of the distance from the origin , that is, the optimal solution of the first Particle position, the position of each particle in the particle swarm is solved, and higher image recognition accuracy and error correction ability can be obtained accordingly, and the best face and body image is reconstructed, face recognition and elevator behavior judgment are performed; The voice perceptual linear prediction model, i.e. a sub-model M3, is used to process the voice into a sequence of voice parameter feature vectors , which contains voice waveform and spectrum, the voice sampling frequency is 10KHz, the frame length is 20ms, so n=200. After the elevator intelligent terminal receives these voice parameters, the current prediction value is calculated according to the following prediction model: ； In the formula: is a prediction coefficient, which is solved by a method of windowing a sequence of speech parameter feature vectors by an autocorrelation method, is a gain factor, which is a mean value of a square of a mean square deviation of the speech parameter feature vectors, is a prediction order, and the greater the value, the closer the prediction model to the original speech signal, and the value is not less than 12, is a prediction error, which has the same frequency as the original speech signal and an amplitude of 25% or less of the amplitude of the original speech signal; and the prediction value is compared with the learning value to identify the speech information result. The elevator safety evaluation model, i.e., the sub-model M4, is an evaluation model of the safety margin and its weight of the parameters affecting the quality of the elevator ride, and the system safety margin is obtained by combining multiple factors such as the running conditions of the system, the design parameters, risk evaluation, etc.: ； In the formula: is the elevator system safety margin, , , is the elevator three-dimensional vector parameter safety margin weight, which is taken as = 0.5, = 0.2, = 0.3, , , is the elevator passenger carrying mass three-dimensional vector parameter national standard value, which is a fixed value under different speeds and different loads, , , is the maximum disturbance value of the elevator passenger carrying mass three-dimensional vector parameter, which is obtained by taking the maximum value of the real-time measurement of the car speed attitude sensor FXAS during the elevator operation; the three-dimensional vector parameters affecting the elevator passenger carrying mass include speed, acceleration, and vibration and car inclination angle; Step 8), the adaptive drive control output is obtained from the model formula (5) a set of adaptive drive control parameters, including proportional coefficient P, integral time I, differential time D; ； ； The system is powered on for the first time to obtain a set of initial control parameters with integral time and differential time being fixed median value 110 ms and proportional coefficient being random value, then the system adjusts dynamically according to the closed-loop feedback result of the elevator load according to the control cycle in the order of proportional coefficient, integral time and differential time, and gradually approaches a set of optimal values, according to the characteristics of the platform elevator load inertia, the adjustment range of the three parameters is: proportional coefficient: 0~100, integral time: 20~200 ms, differential time: 20~200 ms; The initial value is: proportional coefficient: 50, integral time: 80 ms, differential time: 80 ms; The basis of dynamic adjustment is that under the premise of meeting or being better than the national standard value of the elevator load, the proportional coefficient tends to 100, the integral time tends to 20 ms, and the differential time tends to 20 ms, and the system control is in the optimal state; The image recognition accuracy is greater than 95%, and the voice recognition accuracy is greater than 98%; The elevator main control board transmits the proportional coefficient P, integral time I and differential time D drive control parameters to the frequency converter, drives the permanent magnet synchronous motor, and drags the elevator to run, at the same time, every 200 ms, these drive control parameters, elevator state parameters, image and voice interaction information, fault alarm code are uploaded to the remote server to establish the elevator operation database.

Images