Heavy-haul train and longitudinal dynamics traction operation optimization control system and method thereof

A longitudinal dynamics traction operation optimization control system for a heavy-haul train, comprising:

• a motion dynamics model, with control instructions of the train as input, an optimization goal of reducing longitudinal impulse, and desired traction/electrical braking force as output;
• an expert system, with the desired traction/electrical braking force output by the said motion dynamics model, output of an optimization output and feedback module as input, to adjust the desired traction/electrical braking force and feed back adjustment results to the said motion dynamics module;
• a prediction model, with the desired traction/electrical braking force output by the said expert system as input, to set an objective function and predict traction/electrical braking force, wherein
• an expression of the objective function set by the prediction model is:
• min
• ⁢
• J
• ⁡
• (
• k
• )
• =
• ∑
• i
• =
• 1
• p
• q
• i
• [
• F
• p
• (
• k
• +
• i
• )
• -
• F
• r
• (
• k
• +
• i
• )
• ]
• 2
• +
• ∑
• j
• =
• 1
• M
• r
• j
• ⁢
• u
• 2
• (
• k
• )
• ;
• where Fr(k+i) is desired traction/electrical braking force obtained by the said expert system at time k+i, expressed as:
• Fr(k+i)=(1−αi)Fr(k)+αiF(k);
• F(k) is actual traction/electrical braking force at time k;
• αi is a flexibility coefficient computed by the expert system based on train load;
• rj is a weighting coefficient, u(k)=(G1TQG1−R)TGTQ[Fr(k+1)−G2u(k−1)−He(k)], Q=diag[q1, q2, . . . , qp], and q1, q2, . . . , qp are predicted error weighting coefficients;
• R=diag[r1, r2, . . . , rp], and r1, r2, . . . , rp are control quantity weighting matrices;
• H=┌h1, h2, . . . , hp┐r, and h1, h2, . . . , hp are feedback coefficient matrices;
• G=[g1, g2, . . . , gp], g1, g2, . . . , gp are impulse response coefficient matrices, G1 is an impulse coefficient matrix for predicting future conditions, and G2 is an impulse coefficient matrix for past known conditions;
• M is a control time domain length;
• p is a predicted time domain length;
• e(k) is a prediction error at time k, Fp(k)=Fm(k)+He(k);
• e(k)=F(k)−Fm(k);
• Fm(k) is predicted output at time k;
• and
• an optimization output and feedback module, configured to adjust traction/electrical braking force of the train according to the traction/electrical braking force predicted by the prediction model, and feed back the adjusted traction/electrical braking force and real-time monitored coupler force to the expert system.