CN114987434A - Power distribution control method of hybrid power tractor - Google Patents

Abstract

The invention relates to the technical field of vehicle control, in particular to a power distribution control method of a hybrid tractor, which comprises the steps of obtaining historical walking data and operation data of the hybrid tractor; constructing a grey neural network driving speed model prediction model; predicting the current running working condition of the hybrid power tractor based on historical running data and operation data through a grey neural network running speed model to obtain a predicted speed; distributing the walking power and the working power of the hybrid tractor during working by using a discrete random dynamic programming algorithm through a normalized multi-objective optimization function; meanwhile, the torque, the rotating speed, the gear and the air valve opening of an engine and a motor of the hybrid tractor are adjusted by using a discrete random dynamic programming algorithm through a normalized multi-objective optimization function based on the predicted speed, so that the problem that the tractor cannot predict and change an energy supply mode according to a driving state is solved.

Description

Power distribution control method of hybrid power tractor

Technical Field

The invention relates to the technical field of vehicle control, in particular to a power distribution control method of a hybrid tractor.

Background

The development of the second industry is promoted by the intelligent manufacturing technology, and the agricultural industry also urgently needs intelligent, digital and automatic operation machines, so that the effective utilization of resources is realized, the agricultural production efficiency is improved, the agricultural production increase and efficiency increase are promoted, and the income of farmers is changed.

At present, few intelligent control strategies are applied to agricultural machinery equipment, wherein a tractor is taken as an example, the energy supply mode cannot be predicted and changed according to the running state of the tractor, and the power of the tractor during running is large.

Disclosure of Invention

The invention aims to provide a power distribution control method of a hybrid tractor, and aims to solve the problem that the tractor cannot predict and change the energy supply mode according to the driving state.

To achieve the above object, the present invention provides a power distribution control method of a hybrid tractor, comprising the steps of:

acquiring historical walking data and operation data of the hybrid power tractor;

constructing a grey neural network driving speed prediction model;

predicting the current running working condition of the hybrid tractor through the grey neural network running speed prediction model based on the historical walking data and the operation data to obtain a predicted speed;

constructing a normalized multi-objective optimization function;

distributing the walking power and the working power of the hybrid tractor during working by using a discrete random dynamic programming algorithm through the normalized multi-objective optimization function; and simultaneously, based on the predicted speed, using a discrete stochastic dynamic programming algorithm to adjust the torque, the rotating speed, the gear and the air valve opening of the engine and the motor of the hybrid tractor through the normalized multi-objective optimization function.

The specific mode for acquiring the historical walking data and the operation data of the hybrid power tractor is as follows:

searching a MAP (MAP) based on historical working conditions and empirical working conditions;

and acquiring historical walking data and operation data of the hybrid tractor based on the MAP.

The historical walking data comprises the historical running speed of the tractor, wheel required torque, accelerator pedal opening, brake pedal opening, gears and the required power of the whole tractor.

The specific method for constructing the grey neural network driving speed prediction model is as follows:

establishing a grey prediction model and an algorithm mathematical model;

and establishing a grey neural network driving speed prediction model based on the grey prediction model and the algorithm mathematical model.

The method comprises the following steps of predicting the current running working condition of the hybrid tractor through the grey neural network running speed prediction model based on historical walking data and operation data, wherein the specific mode of obtaining the predicted speed is as follows:

grouping the historical walking data according to equal-dimensional recursion to obtain input data and verification data;

training and verifying the gray neural network driving speed prediction model by using the input data and the verification data respectively to obtain an optimal prediction model;

inputting the job data into the optimal predictive model;

the optimal prediction module predicts the current running condition of the hybrid tractor based on the operation data to obtain a predicted speed.

The invention relates to a power distribution control method of a hybrid tractor, which comprises the steps of obtaining historical walking data and operation data of the hybrid tractor; constructing a grey neural network driving speed model prediction model; predicting the current running working condition of the hybrid tractor based on the historical walking data and the operation data through the grey neural network running speed model to obtain a predicted speed; distributing the walking power and the working power of the hybrid tractor during working by using a discrete random dynamic programming algorithm through the normalized multi-objective optimization function so as to realize the comprehensive economy of optimizing the electricity consumption and the oil consumption during running; meanwhile, based on the predicted speed, the torque, the rotating speed, the gear and the air valve opening of the engine and the motor of the hybrid tractor are adjusted by using a discrete random dynamic programming algorithm through the normalized multi-objective optimization function, and the problem that the tractor cannot predict and change the energy supply mode according to the running state is solved.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.

FIG. 1 is a flow chart of a method of controlling power distribution to a hybrid tractor in accordance with the present invention.

Fig. 2 is a flowchart for acquiring historical travel data and work data of the hybrid tractor.

Fig. 3 is a flowchart for constructing a gray neural network travel speed prediction model.

Fig. 4 is a flowchart of the grey neural network driving speed prediction model, which predicts the current driving condition of the hybrid tractor based on the historical traveling data and the operation data to obtain a predicted speed.

Fig. 5 is a diagram of a gray neural network topology.

Detailed Description

Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are illustrative and intended to be illustrative of the invention and are not to be construed as limiting the invention.

Referring to fig. 1 to 5, the present invention provides a power distribution control method of a hybrid tractor, including the steps of:

s1, acquiring historical walking data and operation data of the hybrid tractor;

the concrete mode is as follows:

s11, searching a MAP based on historical working conditions and empirical working conditions;

s12 obtains historical travel data and work data of the hybrid tractor based on the MAP.

Specifically, the historical traveling data comprises the historical traveling speed, wheel required torque, accelerator pedal opening, brake pedal opening, gear and vehicle required power of the tractor.

S2, constructing a grey neural network driving speed prediction model;

the concrete mode is as follows:

s21, establishing a gray prediction model GM (GreyModel) and an algorithm mathematical model;

specifically, the gray prediction model can determine future development trends and provide basis for decision making. The principle is to generate a series of functions through a sequence generation operator, and the functions are the basis of modeling and prediction. The operator can strengthen the uncertainty in the discrete process and weaken the uncertainty.

The neural network is an arithmetic mathematical model for distributed parallel information processing, can process the relationship of each node to achieve the purpose of processing information, and has strong learning ability and adaptability.

The existing working condition data are trained by using a Grey Neural Network (GNN) running speed prediction model, the running speed at 5 moments in the future can be predicted, historical data are updated in real time in the running process of the tractor and retrained, and online real-time predicted running data can be obtained.

Establishing a first-order univariate GM (1,1) model of a gray prediction model

Assuming a time series of length N, a time interval Δ t of 1s, an initial time series X of historical actual speeds ⁽⁰⁾ ，

X ⁽⁰⁾ ＝(X ⁽⁰⁾ (1)，X ⁽⁰⁾ (1)，...，X ⁽⁰⁾ (N))；

To weaken time series X ⁽⁰⁾ The randomness and the volatility of the data letters are accumulated once to generate a new time sequence X ⁽¹⁾ ，

X ⁽¹⁾ ＝(X ⁽¹⁾ (1)，X ⁽¹⁾ (1)，...，X ⁽¹⁾ (N))

＝(X ⁽¹⁾ (1)，X ⁽¹⁾ (1)+X ⁽⁰⁾ (2)，...，X ⁽¹⁾ (N-1)+X ⁽⁰⁾ (N))；

For the time series X ⁽¹⁾ The GM (1,1) model of (1) establishes a differential equation:

where α represents the progression gray number, and u represents the endogenous control gray number.

Wherein the least square method is adopted to pair the parameters

The differential equation is:

can obtain

Wherein the content of the first and second substances,

will be provided with

Introducing differential equation to obtain prediction sequence

Initial time series X ⁽⁰⁾ Is predicted as

To pair

Performing an accumulation subtraction to obtain a predicted sequence

S22, building a grey neural network driving speed prediction model based on the grey prediction model and the arithmetic mathematical model.

The grey model is combined with the neural network model into a Grey Neural Network (GNN) model,

a is a differential equation coefficient, w ═ w ₁ ,w ₂ ,...,w _q ) ^T Is a variable weight matrix;

the predicted sequence expression is:

as shown in fig. 5, a gray neural network topology is constructed and explained per layer.

To a, w _i ，L ₂₁ ,L ₂₂ ,...,L _2n L ₃₁ ,L ₃₂ ,...,L _3n Initializing to obtain u and a node threshold;

each layer outputs the formula:

LA layer: a ═ L ₁₁ t；

LB layer:

LC layer: c. C ₁ ＝bL ₂₁ ,c ₂ ＝x ₁ (t)bL ₂₂ ,...,c _n ＝x _n-1 (t)bL _2n ；

An LD layer: d ═ L ₃₁ c ₁ +L ₃₂ c ₂ +...+W _3n c _n -θX ⁽¹⁾ (0)；

4 historical traveling speeds are input and 1 predicted traveling speed is output at the 2 nd to 5 th nodes of the LC layer nodes.

The predicted running speed of the tractor is obtained through the model.

S3, predicting the current running working condition of the hybrid tractor through the grey neural network running speed prediction model based on the historical walking data and the operation data to obtain a predicted speed;

the concrete mode is as follows:

s31, grouping the historical walking data according to equal-dimensional recursion to obtain input data and verification data;

the historical travel data of the tractor is processed firstly, 30 discrete travel speed data with the interval of 1s are selected, grouping is carried out according to equal dimension recursion according to the travel speed extraction mode of the table 1, and 5 are selected as one group to obtain 20 groups of data. The first 5 of each set are used as input data, denoted as I, and the last 5 are used as verification data, denoted as O, which can be used to verify the accuracy of the model.

When the model is used, the running speed of the tractor 60s is used as a historical database for training, the running speed of the tractor at 5 moments in the future is predicted, and the historical data can be updated in real time in the running process and retrained, so that the model has an online real-time prediction function.

TABLE 1 extraction of travel speed

S32, training and verifying the gray neural network driving speed prediction model by using the input data and the verification data respectively to obtain an optimal prediction model;

s33 inputting the job data into the optimal predictive model;

s34 the optimal prediction module predicts the current running condition of the hybrid tractor based on the operation data to obtain the predicted speed.

S4, constructing a normalized multi-objective optimization function;

the multi-target is processed by the normalization treatment,

the energy management method comprises the following steps of:

L(x)＝Q _D +Q _Y

wherein, g _i Is a weight coefficient, i is 1,2,3,4, Q _Ya ,Q _Da ,Q _Xa ,Q _Za Expresses the oil consumption, the electricity consumption, the walking power and the working power Q after the discrete randomness dynamic programming _Y (r _p ,h _p ),Q _D (r _p ,h _p ),Q _X (r _p ,h _p ),Q _Z (r _p ,h _p ) And the fuel consumption, the power consumption, the walking power and the working power at the moment p are shown, L (x) is the total cost of the power consumption and the fuel consumption, and J is the optimal fuel economy and the optimal power distribution value.

Acquiring the number of ploughshares, the width of the ploughshares, the specific resistance of soil and the tilling depth, further acquiring driving force, traction resistance and rolling resistance, and ensuring that the tractor can normally operate; considering the attachment weight, the structure of a driving device and soil conditions, designing an operation environment for the agricultural machinery based on the operation condition of the agricultural machinery, performing optimization analysis on power distribution by using discrete stochastic dynamic programming, optimizing walking energy consumption and operation energy consumption, solving a multi-objective function by using a normalization algorithm to obtain a global optimal control strategy, and improving the economic fuel property of the agricultural machinery while meeting the operation requirement.

S5, distributing the walking power and the working power of the hybrid tractor during working by using a discrete stochastic dynamic programming algorithm through the normalized multi-objective optimization function; meanwhile, based on the predicted speed, the torque, the rotating speed, the gear and the air valve opening of an engine and a motor of the hybrid tractor are adjusted through the normalized multi-objective optimization function by using a discrete random dynamic programming algorithm;

in particular, the required power of the tractor is obtained,

the traction balance equation of the tractor is as follows: f ═ F _Q +F _z ；

Wherein F represents a driving force, F _Q Indicating tractive effort, F _z Indicating rolling resistance.

Tractive force F _Q Can be expressed as: f _Q ＝mbyδ；

Wherein m represents the number of ploughshares, b represents the width of a single ploughshare, y represents the ploughing depth, and delta represents the soil specific resistance.

Required power for normal operation of the tractor:

wherein, P _req Indicating tractorPower required constantly, F denotes tractor driving force, V _i Indicating the current tractor travel speed, eta _T Indicating mechanical efficiency.

Adjusting the torque, the rotating speed, the gear and the opening of a throttle valve of a motor based on the real-time predicted running speed and the required power of the tractor to obtain the optimal fuel economy;

0.1s is used as a time interval;

decision function h (p) ═ f (T) _e ,i _g )；

Wherein h (p) is a decision function, p represents p time, T _e Representing the torque of the engine, i _g Indicating the speed ratio, T, of the transmission _e 、i _g Are control variables.

The optimization process is discretized into N stages, the state transfer function is,

where f is a system state transition function, r (0) represents an initial value of SOC, r (p) represents an SOC state at time p, and h (p) represents a control amount at time p.

The limiting conditions are as follows:

wherein i _g Representing gear position, T _e 、T _m Respectively representing the torque of the engine, n _e 、n _m Is the rotational speed of the engine and the motor.

In conclusion, the power distribution control method of the hybrid intelligent tractor obtains historical walking data by obtaining the historical running speed, the wheel required torque, the opening degree of an accelerator pedal, the opening degree of a brake pedal, the gear and the required power of the whole tractor; establishing a grey neural network running speed prediction model based on the historical walking data to identify and predict the current running working condition of the tractor to obtain prediction data; based on the predicted driving data, using a discrete random dynamic programming algorithm to adjust the torque, the rotating speed, the gear and the air valve opening of an engine and a motor of the tractor; and normalizing the walking energy consumption and the energy consumption of the operation part of the tractor based on the obtained torque, rotating speed, gear and air valve opening of the engine and the motor to achieve multi-objective optimization. The problem that the optimal fuel economy cannot be matched when the existing agricultural machinery runs and works is solved.

While the above disclosure describes a preferred embodiment of a method for controlling power distribution in a hybrid tractor, it is not intended to limit the scope of the invention, and persons of ordinary skill in the art will understand that all or a portion of the process flow for implementing the above embodiment may be practiced without departing from the scope of the invention.

Claims (5)

1. A method of controlling power distribution to a hybrid tractor, comprising the steps of:

acquiring historical walking data and operation data of the hybrid power tractor;

constructing a grey neural network driving speed prediction model;

predicting the current running working condition of the hybrid tractor through the grey neural network running speed prediction model based on the historical walking data and the operation data to obtain a predicted speed;

constructing a normalized multi-objective optimization function;

distributing the walking power and the working power of the hybrid tractor during working by using a discrete random dynamic programming algorithm through the normalized multi-objective optimization function; and simultaneously, based on the predicted speed, using a discrete stochastic dynamic programming algorithm to adjust the torque, the rotating speed, the gear and the air valve opening of the engine and the motor of the hybrid tractor through the normalized multi-objective optimization function.

2. The hybrid tractor power distribution control method of claim 1,

the specific mode for acquiring the historical walking data and the operation data of the hybrid power tractor is as follows:

searching a MAP (MAP) based on historical working conditions and empirical working conditions;

and acquiring historical walking data and operation data of the hybrid tractor based on the MAP.

3. The hybrid tractor power distribution control method of claim 2,

the historical walking data comprises the historical running speed of the tractor, wheel required torque, accelerator pedal opening, brake pedal opening, gears and the required power of the whole tractor.

4. The hybrid tractor power distribution control method of claim 3,

the specific method for constructing the grey neural network driving speed prediction model is as follows:

establishing a grey prediction model and an algorithm mathematical model;

and establishing a grey neural network driving speed prediction model based on the grey prediction model and the algorithm mathematical model.

5. A method of controlling power distribution to a hybrid tractor based on dynamic discrete stochastic programming according to claim 4,

the method is characterized in that the current running working condition of the hybrid tractor is predicted by the grey neural network running speed prediction model based on the historical walking data and the operation data, and the specific mode of obtaining the predicted speed is as follows:

grouping the historical walking data according to equal-dimensional recursion to obtain input data and verification data;

training and verifying the gray neural network driving speed prediction model by using the input data and the verification data respectively to obtain an optimal prediction model;

inputting the job data into the optimal predictive model;

the optimal prediction module predicts the current running condition of the hybrid tractor based on the operation data to obtain a predicted speed.

Images