Energy management and control method for complex working conditions of mine vehicles

Abstract

The present application relates to the technical field of energy management and control of mine vehicles, and particularly discloses an energy management and control method for complex working conditions of mine vehicles, which obtains a mine vehicle dynamics model and mine vehicle operation related information; performs global planning according to the mine vehicle dynamics model and mine vehicle operation related information and in combination with a dynamic programming algorithm, to obtain mine vehicle operation global planning control variables; performs local optimization on real-time mine vehicle operation information according to a model predictive control algorithm, to obtain mine vehicle operation local planning control variables; and performs weighted mixed coordination processing on the mine vehicle global planning control variables and the mine vehicle local planning control variables, to obtain actual mine vehicle operation control variables. The energy management and control method for complex working conditions of mine vehicles provided by the present application can realize real-time response of mine vehicles under complex working conditions and integrated consideration of energy management.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mine vehicle energy management and control, and particularly relates to an energy management and control method for mine vehicle complex working conditions. BACKGROUND

[0002] With the development of mine dump trucks towards hybrid and pure electric drive, the energy management and motion control integration technology of mine vehicles under complex working conditions (such as steep slope and heavy load operation) faces severe challenges. In the mining area, how to optimize energy use while ensuring the safety control of the vehicle has become an important goal to improve the energy efficiency and stability of the system. However, the energy management system and control system in the prior art are usually designed separately, and this separation leads to the system being unable to quickly respond to changes in working conditions in actual application, affecting the overall energy efficiency and control performance.

[0003] The energy management and control strategy of mine vehicles in the prior art mainly includes the following types: (1) control based on fixed rules: this control method allocates the energy and control power output of the vehicle through pre-set rules and thresholds. Although this method is simple to operate, it lacks flexibility and cannot adjust the strategy according to dynamic working conditions. For example, when the slope changes sharply or heavy load operation occurs, the fixed rules cannot adapt to complex working condition changes, resulting in decreased energy efficiency and control failure. (2) control based on vehicle dynamics model: this method relies on the physical model of the vehicle to optimize energy distribution and power control by calculating the dynamics of the vehicle. However, the working conditions in the mine are complex and variable, especially under steep slope changes and heavy load operation, and the dynamics model is difficult to quickly respond to sudden working conditions. This method has slow response speed and is prone to uneven energy distribution of the vehicle, increasing fuel consumption or electric energy consumption. (3) method based on predictive control: model predictive control (MPC) is a more advanced control method that adjusts the control input and energy output of the vehicle by predicting short-term working conditions to adapt to dynamic environments. However, the optimization effect of MPC depends on the accuracy of the prediction of short-term future working conditions, and it cannot achieve optimal energy management and control in the global range, especially in the complex dynamic environment of the mine, the accuracy of short-term prediction is insufficient, and the optimal control effect cannot be guaranteed.

[0004] In summary, the energy management and control of existing mine vehicles usually considers energy management and motion control separately, resulting in insufficient coordination between energy management and control during actual vehicle driving. Traditional energy management methods, such as dynamic programming (DP), focus on global energy distribution, which optimizes energy use by planning power output throughout the driving process. However, DP cannot respond to dynamic changes of the vehicle under complex working conditions in real time, especially in the case of steep slope changes, load fluctuations, or sudden deceleration of the vehicle in front, and the local control capability is weak.

[0005] Therefore, how to realize real-time response of mine vehicles under complex working conditions and integrated consideration of energy management has become a technical problem to be solved by those skilled in the art. SUMMARY

[0006] The present application provides an energy management and control method for mine vehicles under complex working conditions, which solves the problem that mine vehicles cannot realize real-time response and integrated consideration of energy management under complex working conditions in the related art.

[0007] As an aspect of the present application, an energy management and control method for mine vehicles under complex working conditions is provided, which comprises:

[0008] Obtaining a mine vehicle dynamics model and mine vehicle operation related information, wherein the mine vehicle operation related information at least includes vehicle load, driving path, driving map data and path node information where load changes occur;

[0009] According to the mine vehicle dynamics model and mine vehicle operation related information, global planning is performed in combination with a dynamic programming algorithm to obtain mine vehicle operation global planning control quantity;

[0010] According to the model predictive control algorithm, real-time operation information of the mine vehicle is locally optimized to obtain mine vehicle operation local planning control quantity;

[0011] According to the mine vehicle global planning control quantity and the mine vehicle local planning control quantity, weighted mixed coordination processing is performed to obtain mine vehicle operation actual control quantity.

[0012] Further, according to the mine vehicle dynamics model and mine vehicle operation related information, global planning is performed in combination with a dynamic programming algorithm, which comprises:

[0013] According to the mine vehicle dynamics model and mine vehicle operation related information, state variables and control variables of the dynamic programming algorithm are determined, wherein the state variables include battery state of charge and current speed of the mine vehicle, and the control variables include engine output power and motor output power;

[0014] Determining the objective function of the dynamic programming algorithm;

[0015] The state variables and the control variables are composed of multiple groups of iteration variables, wherein each group of iteration variables includes one state variable factor and one control variable factor;

[0016] According to the way of polling using multiple groups of iteration variables, the objective function of the dynamic programming algorithm is iteratively optimized to obtain mine vehicle operation global planning control quantity, wherein one group of iteration variables is selected for each iteration optimization.

[0017] Further, the real-time operation information of the mine vehicle is locally optimized according to the model predictive control algorithm to obtain mine vehicle operation local planning control quantity, including:

[0018] The running state information of the mine vehicle in the future N time steps is predicted according to the current running information of the mine vehicle, and the running state information at least includes running speed, battery state information and front vehicle distance information;

[0019] The optimal control input information of the future N time steps is solved by the model predictive control algorithm for each time step;

[0020] The local optimization is performed according to the optimal control input information of the current time step, and the above steps are repeated to obtain the mine vehicle operation local planning control quantity.

[0021] Further, the running state information of the mine vehicle in the future N time steps is predicted according to the current running information of the mine vehicle, including:

[0022] The historical speed and historical slope information of the front vehicle are input into the LSTM network to obtain the predicted speed and predicted slope information of the front vehicle in the future N time steps;

[0023] The running state information of the mine vehicle in the future N time steps is predicted according to the predicted speed and predicted slope information in the future N time steps and the current running information of the mine vehicle.

[0024] Further, the optimal control input information of the future N time steps is solved by the model predictive control algorithm for each time step, including:

[0025] The objective function of the model predictive control algorithm is determined, and the expression of the objective function of the model predictive control algorithm is:

[0026]

[0027] Wherein, γ d represents the distance constraint weight, d front,k represents the front vehicle distance of the current mine vehicle, d limit represents the safety distance limit, a v represents the speed weight of the mine vehicle, β SOC represents the deviation weight of the battery state information, δ P represents the power consumption weight;

[0028] The objective function is optimized according to the preset state variable constraint condition, the preset control variable constraint condition and the front vehicle distance preset condition of the mine vehicle to solve the optimal control input information of the future N time steps.

[0029] Further, the global planning control quantity of the mine vehicle and the local planning control quantity of the mine vehicle are weighted and mixed for coordination processing, including:

[0030] The global planning control quantity of the mine vehicle and the local planning control quantity of the mine vehicle are weighted and mixed for coordination processing, including:

[0031] The weighting mixing weight factor in the weighted mixing strategy is dynamically adjusted according to the real-time working condition, so that the global planning control quantity of the mine vehicle and the local planning control quantity of the mine vehicle reach dynamic balance.

[0032] Further, the global planning control quantity of the mine vehicle and the local planning control quantity of the mine vehicle are weighted and mixed for coordination processing, including:

[0033] When the load of the mine vehicle is less than a preset load threshold, the expression of the weighted mixing strategy is:

[0034] P engine,out =α·P engine,DP +(1-α)·P engine,MPC ,

[0035] P motor,out =α·P motor,DP +(1-α)·P motor,MPC ,

[0036] When the load of the mine vehicle is not less than the preset load threshold, the expression of the weighted mixing strategy is:

[0037] P engine,out =α·P engine,MPC +(1-α)·P engine,DP ,

[0038] P motor,out =α·P motor,MPC +(1-α)·P motor,DP ,

[0039] Wherein, P engine,out represents the actual output power of the engine of the mine vehicle, P motor,out represents the actual output power of the motor of the mine vehicle, P engine,DP represents the global planning output power of the engine of the mine vehicle, P motor,DP represents the global planning output power of the motor of the mine vehicle, P engine,MPC represents the local planning output power of the engine of the mine vehicle, P motor,MPC represents the local planning output power of the motor of the mine vehicle, and α represents the weighting mixing weight factor, and 0≤α≤1.

[0040] Further, when the load of the mine vehicle is less than a preset load threshold, the weighted mixing weight factor in the weighted mixing strategy is dynamically adjusted according to a real-time working condition, including:

[0041] When the distance between the mine vehicle and the preceding vehicle is greater than a preset distance threshold and the fluctuation amplitude of the real-time working condition is within a preset amplitude range, the weighted mixing weight factor is increased to increase the weight of the global planning control quantity of the mine vehicle;

[0042] When the distance between the mine vehicle and the preceding vehicle is less than a preset distance threshold and the fluctuation amplitude of the real-time working condition is not within a preset amplitude range, the weighted mixing weight factor is reduced to increase the weight of the local planning control quantity of the mine vehicle.

[0043] Further, when the load of the mine vehicle is not less than a preset load threshold, the weighted mixing weight factor in the weighted mixing strategy is dynamically adjusted according to a real-time working condition, including:

[0044] The TTC and the warning coefficient are calculated in real time according to the current running state information of the mine vehicle;

[0045] The weighted mixing weight factor is dynamically adjusted according to the TTC, the warning coefficient and the distance between the mine vehicle and the preceding vehicle, and the dynamic adjustment expression of the weighted mixing weight factor is:

[0046]

[0047] wherein K represents the warning coefficient, d front represents the distance between the mine vehicle and the preceding vehicle, d limit represents a safety distance threshold, d b represents a braking distance.

[0048] Further, the TTC and the warning coefficient are calculated in real time according to the current running state information of the mine vehicle, including:

[0049] The TTC and the warning coefficient are calculated according to the running speed of the mine vehicle, the distance between the mine vehicle and the preceding vehicle, the relative speed and the load, respectively, wherein the calculation formula of the warning coefficient is:

[0050]

[0051] wherein K represents the warning coefficient, d front represents the distance between the mine vehicle and the preceding vehicle, d b represents a braking distance, m represents the total mass of the mine vehicle, v self represents the running speed of the mine vehicle, F brake represents the braking force of the mine vehicle, d w represents a warning distance, d w = d b +dlimit , d limit denotes a safety distance threshold value;

[0052] The calculation formula of the TTC is:

[0053]

[0054] wherein v rel denotes the relative speed of the mine vehicle and the preceding vehicle, v rel = v self - v front , v front denotes the preceding vehicle speed of the mine vehicle.

[0055] The energy management and control method for complex working conditions of mine vehicles provided by the present application combines global dynamic programming with model predictive control, wherein the global dynamic programming can provide global energy management planning for mine vehicles, ensuring optimal energy efficiency in the overall trip, and the model predictive control dynamically adjusts to real-time changes in working conditions through rolling optimization. This energy management and control method for complex working conditions of mine vehicles, through the control strategy of combining global and local, can dynamically adjust the power output of the engine and the motor considering factors such as slope, load, and vehicle distance, achieving optimal fuel consumption and driving safety, so that the mine vehicle can not only optimize energy use in the global range, but also flexibly respond to unexpected situations in actual working conditions, improving the response speed and flexibility of mine vehicle control. BRIEF DESCRIPTION OF DRAWINGS

[0056] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, and are used together with the following detailed description to explain the present application, but do not constitute a limitation on the present application.

[0057] Figure 1 The flowchart of the energy management and control method for complex working conditions of mine vehicles provided by the present application.

[0058] Figure 2 The flowchart of the global planning provided by the present application.

[0059] Figure 3 The flowchart of the local planning provided by the present application.

[0060] Figure 4 The flowchart of the global and local weighted mixing provided by the present application.

[0061] Figure 5a The effect simulation effect diagram of the iterative dynamic programming IDP and the traditional dynamic programming DP of the present application.

[0062] Figure 5bA comparison chart of fuel consumption and battery state based on the combination of IDP and MPC of the present application and simulation effect of fuel consumption and battery state of MPC alone. DETAILED DESCRIPTION

[0063] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0064] In order for those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should belong to the scope of protection of the present application.

[0065] It should be noted that the terms "first", "second" and the like in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily limit to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0066] An energy management and control method for complex working conditions of a mine vehicle is provided in the embodiment, Figure 1 A flowchart of the energy management and control method for complex working conditions of a mine vehicle provided according to the embodiment of the present application is shown in Figure 1 as shown, comprising:

[0067] S100, obtaining a mine vehicle dynamics model and mine vehicle operation related information, the mine vehicle operation related information at least including vehicle load, driving path, driving map data and path node information where load changes occur;

[0068] In the embodiment of the present application, the mine vehicle dynamics model can specifically include engine output power MAP chart of the mine vehicle, motor output power MAP chart of the mine vehicle and working efficiency of the motor and engine of the mine vehicle, etc.

[0069] S200. Based on the mining vehicle dynamics model and relevant information on mining vehicle operation, and combined with dynamic programming algorithm, global planning is performed to obtain the global planning control quantity for mining vehicle operation.

[0070] In this embodiment of the invention, global dynamic planning can be performed based on the dynamic model of the mining vehicle and relevant information on the operation of the mining vehicle. This global dynamic planning can provide global energy management planning for the mining vehicle, ensuring optimal energy efficiency throughout the entire journey, and thereby obtaining the global planning control quantity for the operation of the mining vehicle.

[0071] It should be understood that the overall planning and control quantities for mining vehicle operation can specifically include the overall engine output power and the overall motor output power for mining vehicle operation.

[0072] S300. Based on the model predictive control algorithm, perform local optimization on the real-time operation information of mining vehicles to obtain the local planning control quantity for the operation of mining vehicles.

[0073] In this embodiment of the invention, the real-time operating information of mining vehicles is locally optimized through model predictive control algorithm. Specifically, it can be understood as making dynamic adjustments to the real-time changes in working conditions through rolling optimization, so as to be able to control the mining vehicles in a timely manner when encountering rapid changes in slope, load fluctuations, or sudden deceleration of the vehicle in front.

[0074] S400. The mining vehicle's global planning control quantity and local planning control quantity are weighted and mixed for coordination processing to obtain the actual control quantity of the mining vehicle's operation.

[0075] It should be understood that by weighting and coordinating the global planning control variables with the local planning control variables, energy use can be optimized globally, and unexpected situations under actual working conditions can be flexibly addressed.

[0076] In summary, the energy management and control method for complex operating conditions of mining vehicles provided by this invention combines global dynamic programming with model predictive control. Global dynamic programming provides a global energy management plan for the mining vehicle, ensuring optimal energy efficiency throughout the entire journey, while model predictive control dynamically adjusts to real-time changes in operating conditions through rolling optimization. This energy management and control method for complex operating conditions of mining vehicles, through a combined global and local control strategy, can dynamically adjust the power output of the engine and motor, considering factors such as gradient, load, and distance, to achieve optimal fuel consumption and driving safety. This allows mining vehicles to optimize energy use globally and flexibly respond to unexpected situations in actual operating conditions, improving the response speed and flexibility of mining vehicle control.

[0077] In this embodiment of the invention, global planning is performed based on the mining vehicle dynamics model and relevant information about mining vehicle operation, combined with a dynamic programming algorithm. Figure 2 As shown, it includes:

[0078] S210. Determine the state variables and control variables of the dynamic programming algorithm based on the dynamic model of the mining vehicle and the relevant information of the mining vehicle operation. The state variables include the battery charging state and the current speed of the mining vehicle, and the control variables include the engine output power and the motor output power.

[0079] It should be noted that, in this embodiment of the invention, the state variables of the dynamic programming specifically refer to the Iterative Dynamic Programming (IDP) algorithm may include:

[0080] SOC (State of Charge): Used to measure the remaining charge of the battery, with a value range of SOC∈[0.3,0.8].

[0081] Current speed of mining vehicles v: Used to describe the current speed of mining vehicles, with a value range of v∈[5,55]km / h;

[0082] Slope θ: Used to describe the slope of the current road, in radians (rad), and is a known constant in the IDP algorithm.

[0083] Control variables include:

[0084] Engine power P engine The power output provided by the engine, with a value range of P. engine ∈[0,1200]kW;

[0085] Motor power P motor The power output provided by the motor has a value range of P. motor ∈[-880,1400]kW. Positive power indicates that the motor provides driving force, while negative power indicates that the motor recovers energy during braking.

[0086] S220. Determine the objective function of the dynamic programming algorithm;

[0087] It should be understood that the goal of the IDP algorithm is to minimize fuel consumption and battery state fluctuations throughout the entire driving path, while maintaining the vehicle speed within a reasonable range. The objective function of the DP algorithm can be expressed as:

[0088]

[0089] Where: D represents the path distance from the starting point to the ending point; α, β, γ, and δ all represent the weight coefficients of different terms; Fueld This represents the fuel consumption of the d-th route segment; SOC d Indicates the battery state of the d-th path segment; v d This represents the vehicle's speed on the d-th segment of the path. The objective function of this IDP algorithm aims to simultaneously optimize factors such as fuel consumption, battery status, and speed stability to ensure that the vehicle achieves overall optimal performance throughout the entire driving process.

[0090] S230. The state variables and the control variables are combined into multiple sets of iterative variables, wherein each set of iterative variables includes a state variable factor and a control variable factor.

[0091] S240. The objective function of the dynamic programming algorithm is iteratively optimized by using multiple sets of iterative variables in a polling manner to obtain the global planning control quantity for the operation of mining vehicles, wherein a set of iterative variables is selected in each iteration optimization.

[0092] It should be understood that, in order to improve computational speed, the IDP algorithm only uses one state variable factor and one control variable factor in each iteration when calculating the objective function. For example, in the current iteration, SOC and engine power can be selected as the iteration variables, while in the next iteration, the current speed of the mining vehicle and motor power can be selected. This approach effectively improves the computational speed of the dynamic programming algorithm by reducing the dimensions of the state space and control space, thereby improving the real-time performance of the dynamic programming algorithm. Specifically, by decomposing the high-dimensional optimization problem into a sequence of low-dimensional subproblems, the dynamic programming in this embodiment actually refers to IDP (Iterative Dynamic Programming), which can significantly reduce computational complexity while ensuring global optimality. Compared with traditional dynamic programming methods, IDP can solve problems quickly under complex conditions while maintaining accuracy, thus meeting the real-time computational needs of the mining transportation system.

[0093] Traditional dynamic programming suffers from exponentially increasing computational complexity due to the discretization requirements of multidimensional state spaces. This invention employs Iterative Dynamic Programming (IDP), which achieves theoretical innovation through a three-layer architecture of "decomposition-cooperation-iteration." Its core idea is to transform a high-dimensional optimization problem into an iterative solution process of multiple low-dimensional subproblems, gradually approaching the global optimum through dynamic information interaction. Compared to traditional DP methods, IDP addresses the issue of excessive computational complexity and achieves a balance between real-time performance and global optimization.

[0094] The IDP algorithm mechanism is primarily reflected in its dimensional decoupling and step-by-step optimization strategy. In the first iteration, a fixed vehicle speed reference trajectory is used, with State of Charge (SOC) as the single state variable to optimize the power distribution between the engine and battery. Next, based on the updated SOC trajectory, vehicle speed is used as the state variable for re-optimization, adjusting the torque output of the drive motor. This step-by-step optimization reduces the computational complexity from the cubic level of traditional DP to the quadratic level, significantly improving computational efficiency. Furthermore, IDP introduces a dynamic mesh adjustment mechanism. In the first iteration, coarse-grained discretization is used to quickly generate an approximate optimal trajectory. Then, based on the trajectory's curvature characteristics, the mesh is refined in key areas (such as rapidly changing SOC segments), and redundant mesh points in low-sensitivity areas are removed based on historical error data. This method concentrates computational resources on high-value state space regions, further improving computational efficiency and optimization accuracy.

[0095] Therefore, IDP effectively solves the core contradiction of traditional DP methods in energy management of mining dump trucks. The iterative optimization process of IDP not only retains the forward-looking advantages of global planning but also integrates a dynamic correction mechanism into the real-time control closed loop, enabling global optimization and local real-time response to work in coordination. This provides an effective collaborative solution for long-term optimization and short-term response. This integrated optimization method not only improves the system's real-time adaptability but also avoids the disconnect between global planning and local control in traditional methods, achieving more efficient energy management.

[0096] The iterative solution process of IDP involves alternating between assumptions and optimizations to progressively determine the optimal control strategy for the system. This process is divided into four main steps, each equivalent to solving a dynamic programming problem with a single state variable and a single control variable. Computational efficiency is significantly improved in IDP because it avoids the complexity of simultaneously handling multiple state and control variables. Compared to traditional dynamic programming methods, IDP demonstrates a clear advantage in computational time.

[0097] The iterative solution process of IDP involves alternating between assumptions and optimizations to progressively determine the optimal control strategy for the system. This process is divided into four main steps, each equivalent to solving a dynamic programming problem with a single state variable and a single control variable. Computational efficiency is significantly improved in IDP because it avoids the complexity of simultaneously handling multiple state and control variables. Compared to traditional dynamic programming methods, IDP demonstrates a clear advantage in computational time.

[0098] Step 1: Assume the vehicle speed sequence for future operating conditions.

[0099] In the first step of IDP, we assume the vehicle speed sequence {V} within the future driving cycle. tThe power sequence {P} is known. This assumption provides the basis for subsequent energy management optimization. Combining the known vehicle speed information and dynamic model, the power sequence {P} of the drive motor can be calculated. edrive,t This provides a basis for subsequent energy management decisions. The specific process is as follows:

[0100] Given the vehicle speed V t The vehicle's mass m and the resistance F during its movement. resist,t The drive motor power P can be calculated using vehicle dynamics equations. edrive,t :

[0101]

[0102] in, F represents the rate of change of vehicle speed. resist,t The drag is caused by road and air resistance.

[0103] Next, based on the vehicle's dynamic model and the known V... t and P edrive,t The SOC (State of Charge) sequence of a battery can be solved using the energy balance equation. t and battery power P bat，t :

[0104]

[0105] Where ΔT is the time step, Q bat For battery capacity, U bat It is the battery's operating voltage.

[0106] This step yields a single state quantity (SOC) and a single control quantity (battery power P). bat,t This dynamic programming problem provides a foundation for subsequent optimization.

[0107] Step 2: Calculate vehicle speed and drive motor power based on SOC and battery power.

[0108] In the first step, the SOC sequence and battery power sequence have been obtained. The next step, the second step, is to recalculate the vehicle's speed sequence {V} based on these results. t} and drive motor power sequence {P edrive,t}

[0109] The specific steps are as follows:

[0110] Given the SOC sequence {SOC t} and battery power sequence {P bat,tThe vehicle speed and drive motor power can be recalculated using the relationship between battery power and drive motor power through vehicle dynamics equations.

[0111]

[0112] Among them, P edrive,t =P engine,t +P bat,t This indicates that the power of the drive motor consists of the sum of the engine power and the battery power.

[0113] This step is also a single state variable (vehicle speed V). t ) and a single control quantity (drive motor power Pedrive) t The dynamic programming problem.

[0114] Step 3: Calculate SOC and battery power based on the new vehicle speed and drive motor power.

[0115] In the second step, the vehicle speed {V} was recalculated. t} and drive motor power {P edrive,t The third step aims to recalculate the SOC and battery power based on this new data. The specific process is as follows:

[0116] Given the new vehicle speed V t and the new drive motor power P edrive,t The new SOC state can be calculated using the state transition equation:

[0117]

[0118] This process also involves a single state quantity (SOC) and a single control quantity (battery power P). bat,t The dynamic programming problem.

[0119] Step 4: Calculate vehicle speed and drive motor power based on SOC and battery power.

[0120] In the third step, a new SOC sequence {SOC} was obtained. t} and battery power sequence {P bat,t The final step is to use this new SOC and battery power to recalculate the vehicle speed V. t and drive motor power P edrive,t The specific steps are as follows:

[0121] Given SOC t and battery power P bat,t The vehicle speed and drive motor power can be recalculated using the vehicle dynamics equations:

[0122]

[0123] This process completes a full iterative cycle, re-optimizing the control strategy for each time step.

[0124] Therefore, the dynamic programming method used in this embodiment of the invention is real iterative dynamic programming. The iterative solution process involves alternating between assumptions and optimizations to gradually approach the optimal control strategy. Through these four iterative steps, IDP solves a simplified dynamic programming problem in each round of calculation, avoiding the complexity of directly handling multiple states and control variables. This step-by-step optimization process significantly reduces computational complexity, making IDP an efficient method for handling complex multi-stage decision problems. Compared with traditional dynamic programming methods, IDP significantly reduces computation time. Assuming that SOC and vehicle speed are discretized into 50 points each, and battery power and drive motor power are also discretized into 50 points, in traditional dynamic programming, the computational load would be 50 × 50 × 50 × 50 = 6,250,000 calculations. However, in IDP, since each step only handles a single state variable and a single control variable, the computational load is only 4 × 50 × 50 = 10,000 calculations, significantly reducing the computational load and greatly improving computational efficiency.

[0125] Specifically, the iterative optimization method includes recursively solving in reverse according to preset initialization endpoint conditions to obtain the global planning control quantity for mine vehicle operation, or calculating in forward according to the control quantity of each stage to obtain the global planning control quantity for mine vehicle operation.

[0126] When implementing iterative optimization, the state transition equation of the IDP algorithm is first determined. Specifically, the speed update equation is determined by the vehicle dynamics equation:

[0127]

[0128] Among them, v new The value represents the velocity at the next time step; ΔT represents the time step size; m represents the total mass of the mining vehicle (including vehicle mass and payload); F d This represents the total resistance of mining vehicles, including air resistance, rolling resistance, and the effect of gradient: Among them, C d The air drag coefficient is represented by A, the frontal area of ​​the vehicle is represented by ρ, the air density is represented by g, and the acceleration due to gravity is represented by f. r θ represents the rolling resistance coefficient, and θ represents the slope angle.

[0129] The SOC update equation is as follows: SOC (State of Battery) is updated using the following formula:

[0130]

[0131] Where: P bat Indicates battery power; Q bat Indicates battery capacity; U bat The battery voltage can be represented by a cubic polynomial fitting model of SOC: U bat =269.7·SOC 3 -512.6 SOC 2 +339.5·SOC+1197.

[0132] Therefore, the optimization process of the IDP algorithm can obtain the optimal power allocation strategy through either reverse or forward solving.

[0133] As a specific implementation method, IDP solves for the optimal power allocation strategy recursively, using the following steps:

[0134] Initialize endpoint conditions: At the end point of the vehicle's journey, set the target battery SOC value and the desired speed as boundary conditions, typically set to SOC. end =0.5, v end =30 / 3.6m / s.

[0135] State transition and recursion: Starting from the endpoint, recursively process in reverse step by step, calculate the state (including SOC and velocity) of each stage through the state transition equation, and select the optimal control input for each stage according to the objective function.

[0136] Cost function calculation: At each time step, IDP calculates the cost corresponding to different control inputs (engine power, motor power) and retains the control strategy with the minimum cost.

[0137] As another specific implementation method, the optimal strategy is solved in the forward direction: after obtaining the optimal control input for each stage, the IDP is calculated in the forward direction to obtain the optimal control strategy and state sequence for the entire driving path.

[0138] Therefore, in this invention, IDP is applied to the global energy management of mining dump trucks. Due to the complexity of mining conditions (such as slope, load, and the influence of preceding vehicles), the global planning of IDP ensures that the vehicle can achieve optimal fuel consumption and maximize energy efficiency throughout the entire driving process. In addition, embodiments of this invention can also combine LSTM to predict preceding vehicle information and slope changes, thereby enabling IDP to rationally plan the power distribution of the entire path, thus providing a global reference for MPC (Model Predictive Control), allowing MPC to perform more precise real-time control under local operating conditions.

[0139] Specifically, the energy management of this IDP during gradient changes can be manifested in the following ways: When going uphill or downhill, the IDP rationally plans the power output of the engine and motor to ensure optimal energy efficiency for the vehicle under different gradients. By considering the impact of gradient, the IDP prioritizes the use of electric motor auxiliary drive to reduce fuel consumption when going uphill, and utilizes electric motor braking to recover energy and reduce battery consumption when going downhill. Control optimization during heavy-load operation: When the vehicle load increases, the IDP adjusts the power distribution between the engine and motor to ensure optimal fuel consumption under heavy load conditions. Simultaneously, the IDP balances changes in battery SOC to avoid excessive consumption of the battery or fuel.

[0140] In this embodiment of the invention, the real-time operation information of mining vehicles is locally optimized based on a model predictive control algorithm to obtain the local planning control variables for the operation of mining vehicles, such as... Figure 3 As shown, it includes:

[0141] S310. Based on the current operating information of the mining vehicle, predict the operating status information of the mining vehicle in the next N time steps. The operating status information includes at least the operating speed, battery status information and the distance to the vehicle in front.

[0142] It should be noted that Model Predictive Control (MPC) primarily optimizes control inputs in real time to achieve the desired system state by predicting the system's behavior over a future period. In this embodiment of the invention, MPC is used to optimize the local energy distribution of mining vehicles. Through rolling time-domain prediction, MPC can adjust the power output of the engine and motor in real time based on the vehicle's current state and future predictions, ensuring optimal energy efficiency and driving safety.

[0143] Specifically, based on the current operating information of the mining vehicles, the operating status information of the mining vehicles in the next N time steps is predicted, including:

[0144] 1) Input the historical speed and historical gradient information of the preceding vehicle into the LSTM network to obtain the predicted speed and predicted gradient information of the preceding vehicle in the next N time steps;

[0145] 2) Based on the predicted vehicle speed and gradient information in the next N time steps, as well as the current operating information of the mining vehicles, predict the operating status information of the mining vehicles in the next N time steps.

[0146] It should be understood that MPC, by combining the speed and distance information of the vehicle ahead predicted by LSTM, can dynamically adjust the power output according to the behavior of the vehicle ahead, ensuring safe following distance while achieving optimal energy efficiency. For example, when the distance to the vehicle decreases, MPC will reduce the vehicle's power output to decelerate, while maintaining energy efficiency as much as possible.

[0147] It's important to note that LSTM (Long Short-Term Memory) is a recurrent neural network (RNN) specifically designed for processing time-series data. It effectively captures long- and short-term dependencies in the data, making it particularly suitable for time-series prediction problems. Unlike traditional RNNs, LSTM overcomes the vanishing and exploding gradient problems that traditional RNNs often encounter with long sequences by introducing memory units and gating mechanisms, enabling it to handle dependencies spanning long periods.

[0148] In this embodiment of the invention, the application of LSTM is mainly reflected in the prediction of the future speed and gradient of the vehicle ahead. By learning from historical vehicle speed and gradient data, LSTM can predict the state of the vehicle ahead at multiple future time steps, helping MPC (Model Predictive Control) to perform more accurate real-time optimization.

[0149] Specifically, the LSTM network structure consists of three main parts:

[0150] Input gate: determines the magnitude of the impact of the current input information on the state of the memory cell;

[0151] Forget gate: determines how much information from a previous state in a memory cell needs to be forgotten;

[0152] Output gate: determines the output at the current moment and updates the state of the memory cell.

[0153] For controlling mining vehicles, the LSTM network takes as input historical vehicle speed and gradient information as input, and outputs predicted vehicle speed and gradient for multiple future time steps. The following is a detailed description of the network structure:

[0154] Input layer: The input consists of three dimensions: the historical speed of the vehicle in front, the historical gradient data, and the predicted signal for the future moment.

[0155] Hidden layers: The network contains two LSTM layers to extract time-series features from the input data;

[0156] Fully connected layer: Features extracted by the LSTM layer are input into the fully connected layer for further processing;

[0157] Output layer: Outputs the predicted speed of the vehicle ahead at multiple future time steps.

[0158] In this embodiment of the invention, the training process of LSTM is a supervised learning process based on time series data. The training data typically includes the input sequence and its corresponding target output. The purpose of training is to continuously adjust the network weights so that LSTM can correctly predict future states based on the input sequence. The detailed training process of LSTM is as follows:

[0159] (1) Data Preparation. Input Data: Historical vehicle speed and historical and future gradient data of the preceding vehicle, with a time step of length t-1. Output Data: The target value used for training is the vehicle speed (and possible gradient) of the preceding vehicle in time steps t, t+1, ..., t+N. In this invention, the input of the LSTM is the speed and gradient data of the preceding vehicle in the past 20 time steps, and the output is the speed prediction value for the next 20 time steps. Constructing the Cell Array of Input Data: Since the LSTM needs to process time series data, the input data needs to be converted into the form of a cell array. For each time step t, the input data consists of the past speed and gradient sequence of the preceding vehicle, and the output is the vehicle speed in the next t+1 to t+N time steps.

[0160] (2) Network Architecture Design: The LSTM network architecture needs to be designed according to the complexity of the time series data. Generally, the more network layers and hidden units there are, the stronger the model's fitting ability, but it is also prone to overfitting. In this embodiment of the invention, the LSTM network contains two LSTM layers, with the first layer having 400 hidden units and the second layer having 200 hidden units. The final output layer is a fully connected layer, outputting the predicted vehicle speed for the next 20 time steps.

[0161] (3) Training process: The LSTM is trained using backpropagation and an optimization algorithm. In this embodiment of the invention, the Adam optimization algorithm is chosen because it has strong convergence and adaptability to dynamic changes. During the training process, the weights and biases in the LSTM network are updated step by step according to the input-output error to minimize the prediction error.

[0162] Therefore, in this embodiment of the invention, LSTM is mainly used to assist MPC in predicting the state of the vehicle ahead. The specific applications of LSTM are reflected in the following aspects:

[0163] Forward vehicle speed prediction: By learning from the historical speed and gradient of the vehicle in front, LSTM can predict the speed of the vehicle in front within a certain number of time steps in the future. This is crucial for the control of MPC, as MPC relies on the prediction of future states to make real-time adjustments during rolling optimization.

[0164] Slope prediction: In addition to speed prediction, LSTM can also predict future slope changes based on historical slope information. This allows the control to adapt more accurately to changes in mining conditions, especially in scenarios where the slope changes frequently.

[0165] The combination of MPC and LSTM: The prediction results of LSTM (speed and gradient at future time steps) serve as input to MPC, helping MPC to consider future speed changes during the optimization process, thereby achieving more flexible and precise control. When the gradient changes drastically, the advance prediction of LSTM can effectively prevent control lag, improving the control's response speed and energy efficiency.

[0166] Solving the multi-input single-output problem: LSTM uses cell arrays to handle multiple inputs (historical speed and slope) and processes them through a time-series recursive network mechanism, outputting speed and slope predictions for a future time period. This structure can effectively solve nonlinear multi-input problems under complex operating conditions.

[0167] S320. For each time step, the optimal control input information for the next N time steps is solved using a model predictive control algorithm.

[0168] It's important to note that the core of MPC (Multi-Process Control) is to use the vehicle's dynamics model to predict future system states and optimize control inputs to ensure that the control reaches the target state within future time steps. At each time step, MPC predicts the system behavior for the next N steps based on the current system state and optimizes control inputs (such as engine power and motor power) to ensure that the vehicle's speed, battery SOC (State of Charge), and other states meet the desired targets.

[0169] In this embodiment of the invention, the state variables of MPC include:

[0170] SOC (State of Charge): Used to describe the remaining battery charge, with a range of SOC ∈ [0.3, 0.8].

[0171] Vehicle current speed v: describes the current speed of the vehicle, in the range of [5,55] km / h;

[0172] Distance d front The distance between a vehicle and the vehicle in front, used to ensure safe driving.

[0173] The control variables for MPC include:

[0174] Engine power P engine The power output provided by the engine ranges from P. engine ∈[0,1200]kW;

[0175] Motor power P motor The power output provided by the motor ranges from P. motor ∈[-880,1400]kW.

[0176] Specifically, for each time step, the optimal control input information for the next N time steps is solved using a model predictive control algorithm, including:

[0177] 1) Determine the objective function of the model predictive control algorithm. The expression of the objective function of the model predictive control algorithm is as follows:

[0178]

[0179] Where, γ d d represents the weight of the vehicle spacing constraint. front,k d represents the distance between the current mining vehicle and the vehicle in front. limit Indicates the safe distance limit, α v β represents the speed weight of mining vehicles. SOC The deviation weight of battery state information, δ P Indicates power consumption weight;

[0180] It should be understood that the goal of MPC is to optimize control inputs to bring vehicle speed, battery SOC, and distance as close as possible to target values ​​over the next N steps, while minimizing power consumption. This objective function ensures efficient energy management and safe driving under complex operating conditions by considering control deviations in speed, SOC, and distance, as well as power and fuel consumption.

[0181] 2) The objective function is optimized based on the preset state variable constraints, preset control variable constraints, and preset distance conditions of the mining vehicle to solve for the optimal control input information for the next N time steps.

[0182] It should be understood that, in order to ensure the practical feasibility of control, MPC will introduce multiple physical constraints during the optimization process to ensure that state values ​​such as power output and SOC are within a reasonable range.

[0183] (1) The constraints on the power of the engine and the motor are as follows:

[0184] 0≤P engine ≤1200kW

[0185] -880≤P motor ≤1400kW.

[0186] (2) Battery SOC constraint. The battery's SOC needs to be maintained within a reasonable range to avoid over-discharge or over-charge:

[0187] 0.3≤SOC≤0.8.

[0188] (3) The vehicle speed and distance constraints are as follows:

[0189] v min ≤v≤v max ,

[0190] dfront >d limit .

[0191] It should be noted that MPC needs to continuously predict future states in the rolling time domain. The state transition equation is used to describe the evolution of the state over time steps, mainly including the updates of velocity and SOC.

[0192] (1) Velocity update equation, mine vehicle speed v k The update equation is:

[0193]

[0194] Where v represents the current speed; ΔT represents the time step; m represents the vehicle mass; F d This represents the total drag, including air resistance, rolling resistance, and the effect of slope.

[0195] (2) The update equation for SOC is:

[0196]

[0197] Among them, Q bat Indicates battery capacity; U bat Represents the battery voltage, based on a polynomial fitting model of SOC.

[0198] S330. Perform local optimization based on the optimal control input information at the current time step, and repeat the above steps to obtain the local planning control quantity for the operation of mining vehicles.

[0199] In this embodiment of the invention, the MPC is solved using the Sequential Quadratic Programming (SQP) algorithm, which simultaneously considers minimizing the cost function and constraints on the control input. SQP offers the following advantages in solving this problem:

[0200] a1. Ability to handle nonlinear systems: MPC is essentially an optimization problem, and the system control in many practical applications is often nonlinear. SQP, as a method for solving nonlinear optimization problems, can effectively handle the nonlinear constraints and objective functions in MPC. By approximating the nonlinear problem as a series of quadratic programming problems and solving them step by step, SQP can adapt well to complex dynamic systems in reality.

[0201] a2. Adaptive Constraints: SQP can handle various constraints in MPC well, including upper and lower limits of input power, battery SOC limitations, and speed and safety distance requirements. This is crucial for the actual physical constraints in the MPC optimization process, as it ensures that the control input does not exceed the physical limitations of the actual system, thus guaranteeing the feasibility of the solution.

[0202] a3. Fast Convergence: Compared to other nonlinear optimization methods, SQP exhibits faster convergence speed when solving constrained optimization problems. It finds the optimal solution in fewer iterations by progressively linearizing the nonlinear constraints and solving a quadratic programming problem in each iteration. This is particularly important for MPC systems requiring real-time control, as it reduces computation time and improves real-time response capabilities.

[0203] a4. High-precision solution: SQP uses quadratic programming to approximate the objective function, and can converge to a relatively accurate optimal solution after multiple iterations. Since MPC usually involves predicting the state and control inputs at multiple future time steps, using SQP can ensure a high-precision optimal solution, thereby improving control performance.

[0204] a5. Flexibility: SQP is not only applicable to objective functions with quadratic forms (such as quadratic cost functions), but can also be extended to a wider range of nonlinear optimization problems. Its flexibility enables MPC to handle various complex control problems, such as nonlinear cost functions in energy management and complex physical dynamic models.

[0205] a6. Effectively addressing uncertainty: In MPC, the prediction of future states may be affected by various uncertainties (such as changes in the behavior of preceding vehicles, sudden changes in road conditions, etc.). SQP, by iteratively solving for the optimal solution at each step, can quickly adjust the prediction and optimization results of future states, making the control more robust to uncertainty.

[0206] In summary, in this embodiment of the invention, MPC, through rolling optimization and real-time prediction, can cope with complex working conditions in mines (such as slope changes, load fluctuations, and preceding vehicle behavior). Specifically, this is manifested in:

[0207] b1. Energy management of slope changes:

[0208] When the vehicle is going uphill or downhill, the MPC can dynamically adjust the power output of the engine and electric motor based on real-time gradient information. When going uphill, the MPC prioritizes engine drive and reduces fuel consumption by using appropriate electric motor assistance; when going downhill, the MPC uses electric motor braking to recover energy and improve energy efficiency.

[0209] b2. Control optimization under heavy load conditions:

[0210] As the load increases, MPC can optimize and adjust the vehicle's power output in real time to ensure a balance between power and fuel consumption under high loads. By predicting future load changes, MPC can make adjustments in advance to avoid power distribution imbalances.

[0211] b3. Responding to dynamic changes in the vehicle in front:

[0212] MPC combines LSTM-predicted speed and distance information of the vehicle ahead to dynamically adjust power output based on the behavior of the vehicle ahead, ensuring safe following distance while achieving optimal energy efficiency. For example, when the distance to the vehicle decreases, MPC reduces the vehicle's power output to decelerate while maintaining energy efficiency as much as possible.

[0213] In this embodiment of the invention, a weighted mixed coordination process is performed based on the global planning control quantity and the local planning control quantity of the mining vehicle, such as... Figure 4 As shown, it includes:

[0214] S410. Perform a weighted hybrid strategy based on the global planning control quantity and the local planning control quantity of the mining vehicle;

[0215] It should be understood that, in order to better combine global optimization and local optimization, the embodiments of the present invention have designed a weighted hybrid strategy, that is, dynamically adjusting the output power ratio of MPC and IDP according to the current operating conditions.

[0216] Specifically, a weighted hybrid strategy is implemented based on the global planning control quantity and the local planning control quantity of the mining vehicles, including:

[0217] (1) When the load of the mining vehicle is less than a preset load threshold, the expression for the weighted mixing strategy is:

[0218] P engine,out =α·P engine,DP +(1-α)·P engine,MPC ,

[0219] P motor,out =α·P motor,DP +(1-α)·P motor,MPC ,

[0220] (2) When the load of the mining vehicle is not less than a preset load threshold, the expression for the weighted mixing strategy is:

[0221] P engine,out =α·P engine,MPC +(1-α)·P engine,DP ,

[0222] P motor,out =α·P motor,MPC +(1-α)·P motor,DP ,

[0223] Among them, P engine,out P represents the actual output power of the engine of a mining vehicle. motor,out P represents the actual output power of the motor in the mining vehicle. engine,DP P represents the global planned output power of the engine in a mining vehicle. motor,DPP represents the global planned output power of the motor of the mining vehicle. engine,MPC P represents the local planned output power of the engine of a mining vehicle. motor,MPC Let α represent the local planning output power of the motor of the mining vehicle, and let α represent the weighted mixed weight factor, where 0≤α≤1.

[0224] S420. The weighted mixing weight factor in the weighted mixing strategy is dynamically adjusted according to the real-time operating conditions so that the global planning control quantity of the mining vehicle and the local planning control quantity of the mining vehicle achieve a dynamic balance.

[0225] As a specific implementation method, when the load of the mining vehicle is less than a preset load threshold, the weighted mixing weight factor in the weighted mixing strategy is dynamically adjusted according to the real-time operating conditions, including:

[0226] (1) When the distance between the mining vehicle and the vehicle in front is greater than the preset distance threshold and the real-time operating condition fluctuation is within the preset range, the weighted mixed weight factor is increased to increase the weight of the global planning control quantity of the mining vehicle.

[0227] (2) When the distance between the mining vehicle and the vehicle in front is less than the preset distance threshold and the real-time operating condition fluctuation is not within the preset range, the weighted mixed weight factor is reduced to increase the weight of the local planning control quantity of the mining vehicle.

[0228] It should be noted that the weighted hybrid weighting factor α is dynamically adjusted according to the current operating conditions. When the vehicle is far from the vehicle in front and the operating conditions do not change much, the weight of IDP is large, that is, α is close to 1; when the distance to the vehicle in front decreases and the gradient changes significantly, MPC gradually takes over control, α decreases, and the weight of MPC increases.

[0229] Specifically, for long-distance driving: when the vehicle is far from the vehicle in front, it mainly relies on the global planning of the dynamic programming (DP), and α≈1. For short-distance driving: when the vehicle approaches the vehicle in front, to ensure safety, the dynamic programming (MPC) gradually takes over control, α decreases, and it relies more on the local optimization of the MPC. For slope changes: when the road slope changes drastically, the MPC prioritizes optimization adjustments, and the dynamic programming (IDP) provides a global reference. In this case, the dynamic adjustment of α is determined based on the rate of change of the slope.

[0230] It should be understood that in energy management and safety control of mining vehicles, the vehicle's load not only affects energy distribution but also significantly influences the calculation of Time to Collision (TTC) and the warning coefficient. Therefore, in the control strategy combining MPC and IDP, to ensure driving safety and energy efficiency optimization under different load conditions, the dynamic impact of vehicle weight on TTC and the warning coefficient is considered when the load exceeds a preset load threshold (which can be understood as the threshold that affects driving safety).

[0231] Based on this, as another specific implementation, when the load of the mining vehicle is not less than a preset load threshold, the weighted mixing weight factor in the weighted mixing strategy is dynamically adjusted according to the real-time operating conditions, including:

[0232] (1) Calculate TTC and early warning coefficient in real time based on the current operating status information of mining vehicles;

[0233] It should be noted that Time to Collision (TTC): TTC refers to the estimated time required for a collision to occur between the vehicle and the vehicle ahead at the current speed. Warning Factor (K): The warning factor measures the safety of the current distance and TTC, and determines the degree of intervention of the MPC in the IDP control strategy.

[0234] Based on the current operating status information of mining vehicles, TTC and early warning coefficients are calculated in real time, including:

[0235] The TTC and warning coefficient are calculated based on the operating speed of the mining vehicles, the distance to the preceding vehicle, the relative speed, and the load. The formula for calculating the warning coefficient is as follows:

[0236]

[0237] Where K represents the warning coefficient, d front d represents the distance between the mining vehicles and the vehicle in front. b Indicates braking distance. m represents the load capacity of the mining vehicle, v self F represents the speed of mining vehicles. brake d represents the braking force of mining vehicles. w Indicates the warning distance, d w =d b +d limit d limit Indicates the safe distance threshold;

[0238] The formula for calculating TTC is as follows:

[0239]

[0240] Among them, v rel v represents the relative speed between the mining vehicle and the vehicle in front. rel =v self -v front v front This indicates the speed of the vehicle in front of the mining vehicle.

[0241] In this embodiment of the invention, K∈[0,1] represents the safety factor. When K is small, it means that the distance between vehicles is too small, and MPC will take over control to increase safety.

[0242] (2) The weighted mixed weighting factor is dynamically adjusted based on TTC, warning coefficient, and distance to the vehicle ahead. The dynamic adjustment expression for the weighted mixed weighting factor is as follows:

[0243]

[0244] Where k represents the warning coefficient, d front d represents the distance between the mining vehicles and the vehicle in front. limit d represents the safe distance threshold. b Indicates braking distance.

[0245] It should be noted that the load on mining vehicles directly affects the braking distance d. b and warning distance d w This significantly impacts TTC and the warning coefficient K. Specifically:

[0246] Braking distance d b Braking distance is directly proportional to vehicle weight (m). The heavier the vehicle, the longer the required braking distance. The formula for calculating braking distance is:

[0247]

[0248] Where m is the total mass of the vehicle; v self F is the vehicle's current speed. brake This refers to the vehicle's braking force, which, for energy-saving control purposes, is typically provided by electric motor braking, with a maximum braking force F. brake =P max / v self , where P max It is the maximum power of the motor braking.

[0249] Warning distance d w Warning distance is the braking distance plus a safety margin; it represents the minimum distance required to maintain a safe following distance under current conditions. The formula for calculating warning distance is:

[0250] d w =d b +d limit ,

[0251] Where d limit It is the safe distance threshold.

[0252] As can be seen from the above formula, an increase in vehicle weight will lead to d b and d w This increases the vehicle's weight, thus affecting the warning coefficient K and TTC. When the vehicle is heavier, it requires a longer braking distance, so the MPC adjusts power distribution and distance control based on real-time vehicle weight and TTC calculations.

[0253] Taking vehicle weight into account, the hybrid strategy of MPC and IDP not only needs to optimize energy distribution but also ensure a safe distance between the vehicle and the vehicle in front, especially when the vehicle weight is high. This part of the strategy dynamically adjusts the power output of the engine and electric motor through TTC and a warning coefficient K. The specific steps are as follows:

[0254] Dynamic calculation of TTC and warning coefficient: Based on the vehicle's current status (speed, distance, relative speed, and vehicle weight), TTC and warning coefficient are calculated in real time.

[0255]

[0256] When the value of K or TTC is less than a certain set value, it indicates that the distance between vehicles is extremely small and there is a risk of collision. MPC will take over control completely and forcibly reduce the power output.

[0257] Weighting adjustments for MPC and IDP: When the vehicle weight is large and the TTC is short, the weight α of MPC will increase significantly to prioritize safety control. When the distance between vehicles is too small, MPC gradually takes over control to reduce vehicle speed and increase the distance between vehicles. The dynamic adjustment formula for the weighting factors is:

[0258]

[0259] When α = 1, MPC takes over control completely to ensure safety.

[0260] Therefore, when considering vehicle load, the control output of the combination of MPC and IDP is specifically the final control output determined based on TTC and the warning coefficient, which is a weighted mixture of MPC and IDP:

[0261] P engine,out =α·P engine,MPC +(1-α)·P engine,DP ,

[0262] P motor,out =α·P motor,MPC +(1-α)·P motor,DP ,

[0263] When the distance between vehicles is large enough, the IDP is responsible for providing global energy distribution, while when the distance between vehicles decreases, the MPC gradually takes over, ensuring vehicle safety through real-time optimization.

[0264] It should be understood that in the hybrid strategy of MPC and IDP, electric motor braking is the primary braking method for vehicles, especially when the vehicle weight is high, where braking power becomes a critical factor. To ensure braking safety, the braking power P of the electric motor... brake It needs to be dynamically adjusted according to the vehicle's load. Specifically:

[0265]

[0266] When P brake Less than the maximum braking power P of the motor max (For example, 880kW), electric motor braking can effectively control vehicle speed.

[0267] When P brake If the maximum braking power is exceeded, the vehicle needs to take additional measures (such as brake caliper braking) to reduce the speed.

[0268] In summary, the embodiments of the present invention, by combining MPC and IDP hybrid strategies of TTC and early warning coefficient, can achieve the following advantages under complex operating conditions:

[0269] Real-time safety assurance: By dynamically calculating TTC and warning coefficients, it can make timely safety controls based on real-time vehicle distance and vehicle weight, ensuring driving safety under high load and short distance conditions.

[0270] Flexible energy allocation: The weighted hybrid strategy of MPC and IDP can make corresponding adjustments according to vehicle weight and distance under different operating conditions, which can ensure the best overall energy efficiency and perform local optimization control in emergency situations.

[0271] Dynamic Response: When the distance between vehicles decreases, the vehicle weight increases, or the vehicle in front suddenly decelerates, the power output and vehicle speed are dynamically adjusted through TTC and warning coefficient to ensure that the vehicle can respond to changes in a timely manner and avoid collisions.

[0272] In summary, the control strategy combining MPC and IDP, by considering the impact of vehicle weight on TTC and warning coefficient, ensures the safety and energy efficiency optimization of vehicles under complex mining conditions, and effectively addresses challenges such as gradient changes, heavy load conditions, and vehicle distance fluctuations.

[0273] Furthermore, by combining MPC with IDP, this invention can achieve comprehensive global and local optimization under complex operating conditions, offering the following advantages:

[0274] The combination of global planning and local optimization: IDP provides the globally energy-efficient path planning, while MPC makes local corrections at each time step to ensure the vehicle's real-time response capability under complex operating conditions.

[0275] Adapting to dynamic operating conditions: MPC combines the speed and distance information of the vehicle in front predicted by LSTM to cope with real-time changing operating conditions, such as changes in slope and shortening distance, to ensure safety and energy efficiency.

[0276] Dynamic weight adjustment: By dynamically adjusting the weight allocation of MPC and IDP, global planning and local optimization can be balanced to achieve the optimal control strategy under different operating conditions.

[0277] Consideration of vehicle weight impact: By taking into account changes in vehicle load, the combined strategy of MPC and IDP can dynamically adjust for heavy and light load conditions to ensure optimal energy efficiency of the vehicle under different load conditions.

[0278] This combined strategy enables stable, efficient, and safe driving control in complex mining conditions, and is suitable for heavy-duty vehicles that require complex energy management.

[0279] The following is combined Figure 5a as well as Figure 5b The simulation diagram shown illustrates the effects of the embodiments of the present invention.

[0280] Figure 5a This is a simulation diagram illustrating the effects of iterative dynamic programming (IDP) and traditional dynamic programming (DP) in an embodiment of the present invention. Figure 5a As can be seen, the simulation results of the IDP in this embodiment of the invention are very close to those of the traditional DP in terms of key indicators. The simulation data of the computational efficiency of this embodiment of the invention are shown in Table 1 below. It can be seen that the IDP in this embodiment of the invention effectively improves the computational efficiency while retaining the performance of the key indicators of DP. That is, the traditional DP takes about 760 seconds because it needs to traverse the entire state space and calculate the global optimal solution by recursively calculating in reverse order. However, the iterative DP in this embodiment of the invention decomposes the problem into subproblems with single state variables and single control variables by alternating between assumptions and optimizations, which significantly reduces the computational complexity and takes only about 1.8 seconds, greatly improving the real-time computing capability.

[0281] Table 1 Comparison of computational efficiency and simulation results between IDP and DP

[0282]

[0283] Figure 5b The figure below shows a comparison of the simulation results of fuel consumption and battery status based on the combination of IDP and MPC in this embodiment of the invention with the simulation results of fuel consumption and battery status based on MPC alone. In addition, Table 2 below shows the specific fuel consumption and SOC change data.

[0284] Table 2 Comparison of Fuel Consumption and SOC Changes

[0285]

[0286] according to Figure 5bAs shown in Table 2, IDP & MPC is more fuel-efficient than MPC, while also having a larger SOC margin. This indicates that, in terms of overall fuel consumption, the IDP & MPC fusion strategy can still maintain low fuel consumption by relying on the globally optimal control of the IDP, demonstrating the effectiveness of collaborative optimization control in energy saving. Meanwhile, the SOC (remaining battery charge) trends are similar under both strategies, and the fusion of global and local control systems results in a larger remaining battery charge at the end of the simulation. This suggests that the collaborative optimization control strategy effectively improves battery management and utilization efficiency while ensuring low energy consumption.

[0287] The above analysis demonstrates that the collaborative optimization strategy of global and local control (IDP and MPC) significantly improves the energy efficiency and safety of hybrid mining dump trucks, especially under complex mining conditions. Simulation and experimental verification show that the integrated control system not only performs excellently in optimizing fuel consumption, state of charge (SOC), and vehicle speed, but also responds in real-time to changes in operating conditions, adjusting the control strategy to ensure efficient vehicle operation under varying road conditions. Compared to traditional single control methods, the integrated control strategy possesses stronger adaptability and real-time optimization capabilities, providing a more practical energy management solution.

[0288] In summary, the energy management and control method for complex working conditions of mining vehicles provided by this invention combines iterative dynamic programming (IDP) for global planning with model predictive control (MPC) for local real-time optimization, constructing a dual optimization control method that combines global and local approaches. By using LSTM to predict the future speed and gradient of the preceding vehicle, the MPC module is assisted in performing more precise control optimization, enabling the control method to adapt to complex mining conditions. The MPC, combined with the prediction information output by LSTM, performs real-time correction of the IDP planning, ensuring that energy distribution and driving safety remain optimal under dynamic conditions. Through dynamic programming, global power planning is performed to ensure that the power distribution of the engine and motor achieves the goal of optimal fuel consumption throughout the entire driving path. At the same time, the MPC considers load and gradient during local optimization, introducing power constraints to ensure reasonable power distribution and practical feasibility. Therefore, the specific advantages of this energy management and control method for complex working conditions of mining vehicles are as follows: 1) Combination of global and local optimization: IDP performs global path planning to ensure optimal fuel consumption throughout the journey; MPC optimizes the IDP results in real time during driving to adapt to complex mining conditions, such as changes in slope and distance to the vehicle in front; 2) Real-time response and safety control: MPC, combined with the speed and slope information of the vehicle in front predicted by LSTM, can achieve precise real-time control under different working conditions, improving driving safety and stability; 3) Improved energy efficiency and reduced fuel consumption: Through optimal energy allocation strategies, especially when the slope changes or the load increases, the power output of the engine and motor can be dynamically adjusted, thereby improving energy efficiency and reducing fuel consumption; 4) Intelligent prediction: The LSTM model can predict future working conditions of the vehicle in front based on historical data, significantly improving the level of intelligence in energy management. This energy management and control method for complex working conditions of mining vehicles can be widely applied to heavy-duty vehicles such as mining dump trucks, performing excellently in dealing with complex working conditions, and is especially suitable for scenarios requiring fine energy management and control.

[0289] It is understood that the above embodiments are merely exemplary implementations used to illustrate the principles of the present invention, and the present invention is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A method for energy management and control of mining vehicles under complex operating conditions, characterized in that, include: Obtain the dynamic model of the mining vehicle and related information on the operation of the mining vehicle. The related information on the operation of the mining vehicle includes at least the vehicle load, driving path, driving map data and path node information where the load changes. Based on the aforementioned mining vehicle dynamics model and relevant information on mining vehicle operation, and combined with a dynamic programming algorithm, global planning is performed to obtain the global planning control quantity for mining vehicle operation. Based on the model predictive control algorithm, the real-time operation information of mining vehicles is locally optimized to obtain the local planning control quantity for the operation of mining vehicles. The actual control quantity of mining vehicle operation is obtained by weighted and mixed coordination processing based on the global planning control quantity and the local planning control quantity of mining vehicle.

2. The energy management and control method for complex working conditions of mining vehicles according to claim 1, characterized in that, Global planning is performed based on the aforementioned mining vehicle dynamics model and relevant information about mining vehicle operation, combined with a dynamic programming algorithm, including: Based on the dynamic model of the mining vehicle and relevant information about the operation of the mining vehicle, the state variables and control variables of the dynamic programming algorithm are determined. The state variables include the battery charging state and the current speed of the mining vehicle, and the control variables include the engine output power and the motor output power. Determine the objective function of the dynamic programming algorithm; The state variables and the control variables are combined into multiple sets of iterative variables, wherein each set of iterative variables includes a state variable factor and a control variable factor; The objective function of the dynamic programming algorithm is iteratively optimized by using multiple sets of iterative variables in a polling manner to obtain the global planning control quantity for the operation of mining vehicles, wherein a set of iterative variables is selected in each iteration.

3. The energy management and control method for complex working conditions of mining vehicles according to claim 1, characterized in that, Based on the model predictive control algorithm, local optimization is performed on the real-time operation information of mining vehicles to obtain local planning control variables for mining vehicle operation, including: Based on the current operating information of the mining vehicle, predict the operating status information of the mining vehicle in the next N time steps. The operating status information includes at least the operating speed, battery status information and the distance to the vehicle in front. For each time step, the optimal control input information for the next N time steps is solved using a model predictive control algorithm; Local optimization is performed based on the optimal control input information at the current time step, and the above steps are repeated to obtain the local planning control quantity for the operation of mining vehicles.

4. The energy management and control method for complex working conditions of mining vehicles according to claim 3, characterized in that, Based on the current operating information of the mining vehicles, predict the operating status information of the mining vehicles in the next N time steps, including: The historical speed and gradient information of the preceding vehicle are input into the LSTM network to obtain the predicted speed and gradient information of the preceding vehicle in the next N time steps; Based on the predicted vehicle speed and gradient information for the next N time steps, as well as the current operating information of the mining vehicles, the operating status information of the mining vehicles for the next N time steps is predicted.

5. The energy management and control method for complex working conditions of mining vehicles according to claim 4, characterized in that, For each time step, the optimal control input information for the next N time steps is solved using a model predictive control algorithm, including: The objective function of the model predictive control algorithm is determined, and its expression is as follows: Where, γ d d represents the weight of the vehicle spacing constraint. front,k d represents the distance between the current mining vehicle and the vehicle in front. limit Indicates the safe distance limit, α v β represents the speed weight of mining vehicles. SOC The deviation weight of battery state information, δ P Indicates power consumption weight; The objective function is optimized based on preset state variable constraints, preset control variable constraints, and preset distance constraints between mining vehicles to find the optimal control input information for the next N time steps.

6. The energy management and control method for complex working conditions of mining vehicles according to claim 1, characterized in that, The weighted and mixed coordination processing based on the global planning control quantity and the local planning control quantity of the mining vehicles includes: A weighted hybrid strategy is applied based on the global planning control quantity and the local planning control quantity of the mining vehicles; The weighted mixing weight factors in the weighted mixing strategy are dynamically adjusted according to real-time operating conditions so that the global planning control quantity and the local planning control quantity of the mining vehicle can achieve a dynamic balance.

7. The energy management and control method for complex working conditions of mining vehicles according to claim 6, characterized in that, A weighted hybrid strategy is implemented based on the global planning control quantity and the local planning control quantity of the mining vehicles, including: When the load of the mining vehicle is less than a preset load threshold, the expression for the weighted mixing strategy is: P engine,out =α·P engine,DP +(1-α)·P engine,MPC , P motor,out =α·P motor,DP +(1-α)·P motor,MPC , When the load of the mining vehicle is not less than a preset load threshold, the expression for the weighted mixing strategy is: P engine,out =α·P engine,MPC +(1-α)·P engine,DP , P motor,out =α·P motor,MPC +(1-α)·P motor,DP , Among them, P engine,out P represents the actual output power of the engine of a mining vehicle. motor,out P represents the actual output power of the motor in the mining vehicle. engine,DP P represents the global planned output power of the engine in a mining vehicle. motor,DP P represents the global planned output power of the motor of the mining vehicle. engine,MPC P represents the local planned output power of the engine of a mining vehicle. motor,MPC Let α represent the local planning output power of the motor of the mining vehicle, and let α represent the weighted mixed weight factor, where 0≤α≤1.

8. The energy management and control method for complex working conditions of mining vehicles according to claim 7, characterized in that, When the load of the mining vehicle is less than a preset load threshold, the weighted mixing weight factor in the weighted mixing strategy is dynamically adjusted according to the real-time operating conditions, including: When the distance between the mining vehicle and the vehicle in front is greater than the preset distance threshold and the real-time operating condition fluctuation is within the preset range, a weighted hybrid weight factor is added to increase the weight of the global planning control quantity of the mining vehicle. When the distance between the mining vehicle and the vehicle in front is less than the preset distance threshold and the real-time operating condition fluctuation is not within the preset range, the weighted mixed weight factor is reduced to increase the weight of the local planning control quantity of the mining vehicle.

9. The energy management and control method for complex working conditions of mining vehicles according to claim 7, characterized in that, When the load of the mining vehicle is not less than a preset load threshold, the weighted mixing weight factor in the weighted mixing strategy is dynamically adjusted according to the real-time operating conditions, including: The TTC and early warning coefficient are calculated in real time based on the current operating status information of mining vehicles; The weighted hybrid weighting factor is dynamically adjusted based on TTC, warning coefficient, and distance to the vehicle ahead. The dynamic adjustment expression for the weighted hybrid weighting factor is as follows: Where K represents the warning coefficient, d front d represents the distance between the mining vehicles and the vehicle in front. limit d represents the safe distance threshold. b Indicates braking distance.

10. The energy management and control method for complex working conditions of mining vehicles according to claim 9, characterized in that, Based on the current operating status information of mining vehicles, TTC and early warning coefficients are calculated in real time, including: The TTC and warning coefficient are calculated based on the operating speed of the mining vehicles, the distance to the preceding vehicle, the relative speed, and the load. The formula for calculating the warning coefficient is as follows: Where K represents the warning coefficient, d front d represents the distance between the mining vehicles and the vehicle in front. b Indicates braking distance. m represents the total mass of the mining vehicle, v self F represents the speed of mining vehicles. brake d represents the braking force of mining vehicles. w Indicates the warning distance, d w =d b +d limit d limit Indicates the safe distance threshold; The formula for calculating TTC is as follows: Among them, v rel v represents the relative speed between the mining vehicle and the vehicle in front. rel =v self -v front v front This indicates the speed of the vehicle in front of the mining vehicle.

Images