A method and system for active thermal management control of engineering machinery based on deep learning

Abstract

The present invention provides an active thermal management control method and system for engineering machinery based on deep learning. The present invention uses a support vector machine improved by an artificial rabbit and whale fusion algorithm to train historical data to obtain a fan speed prediction model, and then combines Markov chain and association rule algorithms to construct a weighted operating condition that can more accurately reflect the actual heat dissipation needs of engineering machinery. The collected real-time data and weighted operating conditions are used as inputs of the fan speed prediction model, and the fan speed is output to improve the control strategy of the thermal management system of engineering machinery, so that the system can meet the thermal management needs of engineering machinery while minimizing energy consumption.

Description

Technical Field

[0001] The present invention belongs to the technical field of thermal management control of engineering machinery, and specifically relates to an active thermal management control method and system for engineering machinery based on deep learning. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] In the field of construction machinery, especially in operations involving large or even heavy construction machinery, meeting thermal load requirements has always been an important technical challenge. These construction machinery will generate a lot of heat during long-term operation. If the heat cannot be effectively dissipated, it will cause the construction machinery to overheat, performance degradation or even damage, thus affecting the reliability and service life of the construction machinery.

[0004] Traditional thermal management control methods for construction machinery usually use temperature monitoring and control based on empirical rules or simple physical models, but this method often cannot meet the precise control requirements under complex working conditions. Due to the complexity and variability of the working environment, traditional methods often find it difficult to accurately predict and adjust the thermal management control strategy of construction machinery, which can easily lead to overheating or overcooling, affecting the stability and efficiency of construction machinery. Summary of the invention

[0005] In order to solve the above problems, the present invention proposes an active thermal management control method and system for engineering machinery based on deep learning. The present invention uses a support vector machine improved by a deep learning algorithm to train historical data to obtain a fan speed prediction model, and combines Markov chain and association rule algorithms to construct a weighted working condition that can more accurately reflect the actual heat dissipation needs of engineering machinery, improve the control strategy of the thermal management system of engineering machinery, so that the system can meet the thermal management needs of engineering machinery while minimizing energy consumption.

[0006] According to some embodiments, the present invention adopts the following technical solutions:

[0007] A first aspect of the present invention provides an active thermal management control method for engineering machinery based on deep learning.

[0008] A deep learning-based active thermal management control method for construction machinery comprises the following steps:

[0009] The support vector machine is improved by using a deep learning algorithm to obtain an improved support vector machine algorithm. The improved support vector machine algorithm is used to train historical data samples including ambient temperature, thermal management parameters of construction machinery, powertrain operating parameters and fan speed to form a prediction model for fan speed.

[0010] Obtain the micro-load segment of the current load spectrum, use the Markov chain model to establish the probability transfer matrix of various working conditions, use the association rule algorithm to calculate the support and confidence of the relevant item sets, and construct the association rule score based on the support and confidence, which is integrated with the transfer probability of the corresponding state to predict the next micro-load segment;

[0011] Calculating the powertrain operating parameters of the construction machinery under the weighted working condition based on the weights of the current micro-load segment and the predicted micro-load segment;

[0012] The real-time ambient temperature, thermal management parameters of the construction machinery and the operating parameters of the powertrain under weighted working conditions are used as inputs of the prediction model to predict the fan speed, which is then converted into a control signal.

[0013] As an optional implementation plan for the first aspect, the specific process of improving the support vector machine using a deep learning algorithm includes: introducing the whale surrounding prey and the spiral bubble method attacking prey position update strategy in the whale algorithm into the artificial rabbit algorithm, improving its detour foraging and random hiding behaviors, obtaining an artificial rabbit and whale fusion algorithm, and using the artificial rabbit and whale fusion algorithm to improve the support vector machine.

[0014] The artificial rabbit and whale fusion algorithm specifically includes: the improved detour foraging formula is:

[0015] ;

[0016] ;

[0017] = ;

[0018] ;

[0019] ;

[0020] ;

[0021] ;

[0022] ;

[0023]

[0024]

[0025] ;

[0026] in, For the In the iteration Rabbit candidate position; For the In the iteration The position of the rabbit, For the In the iteration The position of the rabbit; It is the individual dimension of artificial rabbit; represents the running operator; is the mapping vector; Indicates from 1 to A random integer between ; for the running length; is a random number that follows a standard normal distribution; is a random number between (0, 1), The updated position of the artificial rabbit using the whale to surround the prey and the spiral bubble method to attack the prey. is the optimal artificial rabbit position, is the maximum number of iterations, To return the rounded value; is the coefficient vector, is the distance between the artificial rabbit and the random individual position, For the The position of random individuals in the artificial rabbit population at the iteration, is a random number in the range [-1, 1], Convergence factor, is a random vector in [0,1], is a random variable in [0,1], and the position of the artificial rabbit is adjusted using this random variable. is the spiral constant, whose value is 1;

[0027] The improved random hiding formula is:

[0028] ;

[0029]

[0030] ;

[0031] ;

[0032] in The updated position of the artificial rabbit using the whale to surround the prey and the spiral bubble method to attack the prey. For the In the iteration The first rabbit produced a cave; is a hidden parameter; is a random number from a standard normal distribution; , is a random number between (0, 1); is the mapping vector, is the total number of artificial rabbits.

[0033] As an optional implementation scheme for the first aspect, the specific process of improving the support vector machine using the artificial rabbit and whale fusion algorithm includes: setting initial parameters for the support vector machine algorithm improved based on the fusion of the artificial rabbit and whale algorithms, randomly generating a combination of penalty factors and variance parameters of the radial basis kernel function as the initial population, and finding the combination of penalty factors and variance parameters of the radial basis kernel function that minimizes the error of the support vector machine prediction model.

[0034] As an optional implementation scheme of the first aspect, the thermal management parameters of the engineering machinery include engine coolant temperature, hydraulic system coolant temperature and heat dissipation of each heat dissipation module;

[0035] Powertrain operating parameters include engine speed, engine torque, hydraulic cylinder load, and construction machinery travel speed;

[0036] For the thermal management system of construction machinery, different ambient temperatures, engine speeds, engine torques, hydraulic cylinder loads, engine coolant temperatures, hydraulic system coolant temperatures, heat dissipation of the cooling module, and construction machinery travel speeds are set as training sample inputs. The fan speeds that meet the safety temperature requirements of the construction machinery are simulated and used as training samples for the prediction model.

[0037] The improved support vector machine is used to train the samples to form a prediction model for fan speed.

[0038] As an optional implementation scheme of the first aspect, the process of predicting the next micro-load segment using the Markov chain and association rule algorithm includes:

[0039] Based on the micro-load fragment of the current load spectrum, calculate the support of all 2-item sets, the current state Transition to next state , the item set The support and confidence are:

[0040] ;

[0041] ;

[0042] Support Represents all itemsets Likelihood of occurrence, confidence Indicates prerequisites If another transaction occurs The probability of occurrence, P represents the probability, calculated in the current state The maximum support of all item sets under , all 2-itemset supports are normalized, and the normalization formula is:

[0043] ,

[0044] The normalized support and confidence are used to obtain the association rule score using the following function:

[0045] ,

[0046] ,

[0047] ,

[0048] in, It is an item set The support of It is an item set The confidence level, It is an item set Association rule scoring;

[0049] The improved state transition probability is obtained by multiplying the association rule score with the corresponding state transition probability, and the working condition with the maximum improved transition probability is selected as the prediction result of the next state load.

[0050] As an optional implementation scheme of the first aspect, based on the weights of the current micro-load segment and the predicted micro-load segment, the specific process of calculating the powertrain operating parameters of the engineering machinery under the weighted working condition includes:

[0051] The objective weight is determined according to the difference between the load spectra of the two fragments. The weight of the moment is:

[0052] ,

[0053] in , for The load of construction machinery at all times, for The load of construction machinery at all times, for The weight of the moment, is the coefficient, is a constant;

[0054] because Moment and The moment load weight correlation coefficient is -1, and the weighted load is:

[0055] .

[0056] As an optional implementation plan for the first aspect, the real-time ambient temperature, the engine coolant temperature, the hydraulic system coolant temperature, the heat dissipation of the cooling module, the construction machinery's travel speed, and the engine speed, engine torque and hydraulic cylinder load under weighted working conditions are used as inputs to a fan speed prediction model to predict the fan speed, which is then converted into a PWM duty cycle that drives the fan's variable plunger pump speed to form a control strategy.

[0057] A second aspect of the present invention provides an active thermal management control system for engineering machinery based on deep learning.

[0058] An active thermal management control system for construction machinery based on deep learning, comprising:

[0059] A deep learning algorithm improvement module is configured to improve a support vector machine using a deep learning algorithm to obtain an improved support vector machine algorithm, and to train historical data samples including ambient temperature, thermal management parameters of construction machinery, powertrain operating parameters, and fan speed using the improved support vector machine algorithm to form a prediction model for fan speed;

[0060] The micro-load segment prediction module is configured to obtain the micro-load segment of the current load spectrum, establish the probability transfer matrix of various working conditions by using the Markov chain model, calculate the support and confidence of the relevant item sets by using the association rule algorithm, and construct the association rule score based on the support and confidence, and combine it with the transfer probability of the corresponding state to predict the next micro-load segment;

[0061] A weighted operating condition parameter calculation module is configured to calculate a powertrain operating parameter of the construction machinery under a weighted operating condition based on weights of a current microload segment and a predicted microload segment;

[0062] The fan speed prediction module is configured to use the real-time ambient temperature, thermal management parameters and powertrain operating parameters under weighted working conditions as input quantities of the prediction model, predict the fan speed, and convert it into a control signal.

[0063] A third aspect of the present invention provides a computer-readable storage medium.

[0064] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps in the deep learning-based active thermal management control method for engineering machinery as described above.

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] (1) The present invention utilizes the artificial rabbit and whale fusion algorithms to improve the support vector machine to obtain an improved support vector machine. By combining the artificial rabbit and whale fusion algorithms, the performance of the support vector machine can be improved. The improved support vector machine is applied to historical data samples for training, and a more accurate and reliable prediction model can be obtained.

[0067] (2) The present invention provides an active thermal management control method for construction machinery based on deep learning. It uses Markov chain and association rule algorithms to predict the next micro-load segment based on the micro-load segment of the current load spectrum, uses a specific formula to calculate the weights of the existing segment and the predicted segment, introduces the weighted load spectrum into the prediction model to obtain the fan speed, forms an active thermal management control strategy for construction machinery, and combines components such as radiators, fans, and hydraulic systems to form an active thermal management system for construction machinery. It uses a custom objective weighting method formula to construct a weighted working condition that can more accurately reflect the actual heat dissipation requirements of construction machinery on the basis of Markov chain and association rule algorithms for the control of the active thermal management system of construction machinery, making the fan speed prediction more accurate.

[0068] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0070] Figure 1 The flowchart of an embodiment is shown in FIG. DETAILED DESCRIPTION

[0071] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0072] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0073] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0074] In the absence of conflict, the embodiments in this application and the features in the embodiments may be combined with each other.

[0075] The active thermal management system for construction machinery mentioned in the following embodiments includes components such as a radiator, an engine, a hydraulic system, a water pump, a fan and a condenser.

[0076] Embodiment 1

[0077] This embodiment provides an active thermal management control method for construction machinery based on deep learning, which specifically includes the following steps:

[0078] A model for predicting fan speed is proposed based on the support vector machine improved by the artificial rabbit and whale fusion algorithm. The improved support vector machine is improved by using the artificial rabbit and whale fusion algorithm. The improved support vector machine algorithm is used to train historical data samples including ambient temperature, engine coolant temperature, hydraulic system coolant temperature, heat dissipation of cooling module, construction machinery travel speed, engine speed, engine torque, hydraulic cylinder pressure and fan speed, and finally a prediction model for fan speed is formed.

[0079] In this embodiment, the fan speed prediction model is obtained by training and learning historical data samples using a support vector machine algorithm improved by the fusion of an artificial rabbit and a whale algorithm.

[0080] Specifically, the artificial rabbit optimization algorithm uses the foraging and hiding strategies of real rabbits. By switching between the two strategies through energy contraction, the algorithm includes the following three behaviors.

[0081] 1) Take a detour to find food

[0082] When rabbits forage for food, they always look for grass far away and ignore nearby grass. They only eat grass randomly in other areas instead of eating grass in their own area. This foraging behavior is called detour foraging. Its mathematical model is as follows:

[0083]

[0084]

[0085] ;

[0086] ;

[0087]

[0088] ;

[0089]

[0090] ;

[0091] Where: For the In the iteration Rabbit candidate position; For the In the iteration The position of the rabbit, For the In the iteration The position of the rabbit; n is the number of artificial rabbits; represents the running operator; To return the rounded value; is the mapping vector; Indicates from 1 to A random integer between ; for the running length; is a random number that follows a standard normal distribution; is a random number between (0, 1); It is the individual dimension of artificial rabbit; is the maximum number of iterations.

[0092] 2) Random hiding

[0093] In each iteration, a rabbit always generates a hole along the dimension of the search space, and always randomly selects one hole from all the holes to hide in to reduce the probability of being preyed on. When rabbits encounter danger, they randomly select a hole to hide. This random hiding strategy model is as follows:

[0094]

[0095] ;

[0096] ;

[0097] ;

[0098] Where: For the In the iteration The first rabbit produced a cave; is a hidden parameter; is a random number from a standard normal distribution; is a random number between (0, 1); is the mapping vector, is the number of artificial rabbits. After implementing the detour foraging or random hiding strategy, the position of the rabbit is updated as follows:

[0099] .

[0100] Where: is the fitness function used for the optimization problem.

[0101] 3) Energy reduction

[0102] The switch between the detour foraging and random hiding strategies is achieved through energy reduction. Over time, the rabbit's energy will gradually decrease. The energy factor in ARO is defined as follows:

[0103] ;

[0104] In the formula, is a random number between (0, 1). As time increases, the energy factor tends to 0. When , individuals in the population implement the detour foraging strategy; otherwise, individuals in the population implement the random hiding strategy.

[0105] Although ARO outperforms many classic nature-inspired algorithms, it still has some shortcomings, such as lack of population diversity and easy to fall into local optimality.

[0106] This example uses the Whale Algorithm (WOA) to improve the Artificial Rabbit Algorithm. The Whale surrounds the prey and the spiral bubble method is introduced into the Artificial Rabbit Algorithm to improve its detour foraging and random hiding behaviors.

[0107] The improved detour foraging formula is as follows:

[0108] ;

[0109] ;

[0110] = ;

[0111] ;

[0112] ;

[0113] The improved random hiding formula is as follows:

[0114] ;

[0115] in The updated position of the artificial rabbit using the whale-surrounding prey and spiral bubble method to attack prey, is the optimal artificial rabbit position, is the coefficient vector, is the distance between the artificial rabbit and the random individual position, For the The position of random individuals in the artificial rabbit population at the iteration, is the spiral constant, whose value is 1, is a random number in the range [-1, 1], Convergence factor, is a random vector in [0,1], is a [0,1] random variable, which is used to adjust the position of the artificial rabbit.

[0116] By utilizing the fact that the whale's search agent in the whale algorithm moves both in a straight line and in a spiral, with a 50% probability each, the flexibility of searching for the artificial rabbit is improved, which can prevent the artificial rabbit algorithm from falling into the local optimum to a certain extent.

[0117] Thus, the artificial rabbit and whale fusion algorithm is obtained.

[0118] In the process of training the fan speed prediction model, the support vector machine algorithm improved by the fusion of artificial rabbit and whale algorithms sets the initial parameters, randomly generates a combination of penalty factors and variance parameters of radial basis kernel functions as the initial population, and searches for the combination of penalty factors and variance parameters of radial basis kernel functions that minimizes the error of the support vector machine prediction model.

[0119] Of course, in other embodiments, the support vector machine algorithm may be improved by other deep learning algorithms, such as the artificial fish algorithm.

[0120] The training samples of the fan speed prediction model are obtained through the joint operation of one-dimensional and three-dimensional thermal management software simulation. For the thermal management system of construction machinery, different ambient temperatures, engine speeds, engine torques, hydraulic cylinder loads, engine coolant temperatures, hydraulic system coolant temperatures, heat dissipation of the cooling module, and construction machinery travel speeds are set as training sample inputs to simulate the fan speeds that meet the safety temperature requirements of construction machinery, and the above data are used as training samples for the support vector machine prediction model.

[0121] The support vector machine algorithm improved by the artificial rabbit and whale fusion algorithm was trained and learned on the above historical data samples to obtain a prediction model for the fan speed of construction machinery.

[0122] The micro-load segment of the current load spectrum is obtained, and the next micro-load segment is predicted using the Markov chain and association rule algorithm.

[0123] Specifically, before using the Markov chain and association rule algorithm to predict the next micro-load segment, it also includes:

[0124] Based on the existing data, the probability transfer matrix of various working conditions is established. Assume that the current state load is , the next state load is ,make The state load at the moment is , then from the current state Transition to next state The transition probability can be expressed as:

[0125] ;

[0126] in: is the current load Reach next state load The transition probability, is the current load Reach next state load The number of events that occurred, is the current load The total number of occurrences of reaching any next state load.

[0127] Association rules are of the form The implied form of , ,and , As a prerequisite, is the corresponding association result, is an item set.

[0128] The strength of association can be expressed by support and confidence.

[0129] Support represents the value of a certain item set among all item sets. The possibility of occurrence. It can be expressed as follows:

[0130] ;

[0131] Confidence indicates prerequisites If another transaction occurs The probability of occurrence. It can be expressed as follows:

[0132] ;

[0133] In the specific implementation, when using the Markov chain model to predict the next micro-load segment, based on the micro-load segment of the current load spectrum, the support of all 2-item sets is calculated, and the current state Transition to next state For example, the itemset The support and confidence are:

[0134] ;

[0135] ;

[0136] Calculate in current state The maximum support of all item sets under , all 2-itemset supports are normalized, and the normalization formula is:

[0137] ;

[0138] The normalized support and confidence are used to obtain the association rule score using the following function:

[0139] ;

[0140] ;

[0141] ;

[0142] in, It is an item set The support of It is an item set The confidence level, It is an item set Association rule scoring.

[0143] The improved state transition probability is obtained by multiplying the association rule score with the corresponding state transition probability, and the working condition with the maximum improved transition probability is selected as the prediction result of the next state load.

[0144] Specifically, based on the current load , select the working condition with the maximum improved transition probability as the next state load The prediction results are:

[0145] ;

[0146] Finally, the next state load predicted by Markov chain and association rule algorithm is obtained.

[0147] Based on the weights of the current micro-load segment and the predicted micro-load segment, the engine speed, engine torque and hydraulic cylinder load under the weighted working condition are calculated.

[0148] The objective weight is determined according to the difference between the load spectra of the two segments. The greater the difference, The more information provided at all times, the greater the role it can play in the comprehensive evaluation and the greater its weight. The weight of the moment is:

[0149] ;

[0150] in , for The load of construction machinery at all times, for The load of construction machinery at all times, for The weight of the moment, is a coefficient. In this embodiment, its value is 0.4. is a constant. In this embodiment, its value is 27.73.

[0151] because Moment and The moment load weight correlation coefficient is -1, and the weighted load is:

[0152] ;

[0153] The real-time ambient temperature, engine coolant temperature, hydraulic system coolant temperature, heat dissipation of the cooling module, construction machinery travel speed, as well as engine speed, engine torque and hydraulic cylinder load under weighted working conditions are used as inputs of the fan speed prediction model to predict the fan speed. The PWM duty cycle of the variable piston pump driving the fan is converted to obtain the active thermal management control strategy of the construction machinery.

[0154] Based on the obtained active thermal management control strategy for construction machinery, combined with components such as radiators, fans, and hydraulic systems, an active thermal management system for construction machinery is finally formed.

[0155] Embodiment 2

[0156] This embodiment provides an active thermal management control system for engineering machinery based on deep learning, including:

[0157] A deep learning algorithm improvement module is configured to improve a support vector machine using a deep learning algorithm to obtain an improved support vector machine algorithm, and to train historical data samples including ambient temperature, thermal management parameters of construction machinery, powertrain operating parameters, and fan speed using the improved support vector machine algorithm to form a prediction model for fan speed;

[0158] The micro-load segment prediction module is configured to obtain the micro-load segment of the current load spectrum, establish the probability transfer matrix of various working conditions by using the Markov chain model, calculate the support and confidence of the relevant item sets by using the association rule algorithm, and construct the association rule score based on the support and confidence, and combine it with the transfer probability of the corresponding state to predict the next micro-load segment;

[0159] A weighted operating condition parameter calculation module is configured to calculate a powertrain operating parameter of the construction machinery under a weighted operating condition based on weights of a current microload segment and a predicted microload segment;

[0160] The fan speed prediction module is configured to use the real-time ambient temperature, thermal management parameters and powertrain operating parameters under weighted working conditions as input quantities of the prediction model, predict the fan speed, and convert it into a control signal.

[0161] Thermal management parameters of construction machinery include engine coolant temperature, hydraulic system coolant temperature and heat dissipation of each cooling module;

[0162] The powertrain operating parameters include engine speed, engine torque, hydraulic cylinder load and construction machinery travel speed.

[0163] It should be noted here that the various modules in the deep learning-based active thermal management control system for engineering machinery in this embodiment correspond one-to-one to the various steps in the deep learning-based active thermal management control method for engineering machinery in Example 1, and their specific implementation processes are the same, which will not be repeated here.

[0164] Embodiment 3

[0165] This embodiment provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the steps in the deep learning-based active thermal management control method for engineering machinery provided in Embodiment 1 are implemented.

[0166] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0167] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0168] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0169] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0170] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made by those skilled in the art within the spirit and principle of the present invention without creative labor shall be included in the protection scope of the present invention.

Claims

1. A deep learning-based active thermal management control method for construction machinery, characterized in that: The following steps are involved: The support vector machine is improved by using a deep learning algorithm to obtain an improved support vector machine algorithm. The improved support vector machine algorithm is used to train historical data samples including ambient temperature, thermal management parameters of construction machinery, powertrain operating parameters and fan speed to form a prediction model for fan speed. The specific process of using deep learning algorithm to improve support vector machine includes: introducing the whale encircling prey and spiral bubble attack prey position update strategies in the whale algorithm into the artificial rabbit algorithm, improving its detour foraging and random hiding behaviors, obtaining the artificial rabbit and whale fusion algorithm, and using the artificial rabbit and whale fusion algorithm to improve support vector machine; The specific process of improving the support vector machine by using the artificial rabbit and whale fusion algorithm includes: the support vector machine algorithm based on the fusion of the artificial rabbit and whale algorithms sets initial parameters, randomly generates a combination of the variance parameters of the penalty factor and the radial basis kernel function as the initial population, and searches for a combination of the variance parameters of the penalty factor and the radial basis kernel function that minimizes the error of the support vector machine prediction model; Obtain the micro-load segment of the current load spectrum, use the Markov chain model to establish the probability transfer matrix of various working conditions, use the association rule algorithm to calculate the support and confidence of the relevant item sets, and construct the association rule score based on the support and confidence, which is integrated with the transfer probability of the corresponding state to predict the next micro-load segment; Calculating the powertrain operating parameters of the construction machinery under the weighted working condition based on the weights of the current micro-load segment and the predicted micro-load segment; Based on the weights of the current micro-load segment and the predicted micro-load segment, the specific process of calculating the powertrain operating parameters of the construction machinery under the weighted working condition includes: The objective weight is determined according to the difference between the load spectra of the two fragments. The weight of the moment is: ； in , for The load of construction machinery at all times, for The load of construction machinery at all times, for The weight of the moment, is the coefficient, is a constant; because Moment and The moment load weight correlation coefficient is -1, and the weighted load is: ； The real-time ambient temperature, thermal management parameters of the construction machinery and the operating parameters of the powertrain under weighted working conditions are used as inputs of the prediction model to predict the fan speed, which is then converted into a control signal.

2. The method for active thermal management control of construction machinery based on deep learning as claimed in claim 1, characterized in that: The artificial rabbit and whale fusion algorithm specifically includes: the improved detour foraging formula is: ； ； = ； ； ； ； ； ； ； in, For the In the iteration Rabbit candidate position; For the In the iteration The position of the rabbit, For the In the iteration The position of the rabbit, It is the individual dimension of artificial rabbit; represents the running operator; is the mapping vector; Indicates from 1 to A random integer between ; for the running length; is a random number that follows a standard normal distribution; is a random number between (0, 1), The updated position of the artificial rabbit using the whale to surround the prey and the spiral bubble method to attack the prey. is the optimal artificial rabbit position, is the maximum number of iterations, To return the rounded value; is the coefficient vector, is the distance between the artificial rabbit and the random individual position, For the The position of random individuals in the artificial rabbit population at the iteration, is a random number in the range [-1, 1], Convergence factor, is a random vector in [0,1], is a [0,1] random variable, and the position of the artificial rabbit is adjusted using this random variable. is the spiral constant, whose value is 1; The improved random hiding formula is: ； ； ； ； in The updated position of the artificial rabbit using the whale to surround the prey and the spiral bubble method to attack the prey. For the In the iteration The first rabbit produced a cave; is a hidden parameter; is a random number from a standard normal distribution; , is a random number between (0, 1); is the mapping vector, is the number of artificial rabbits.

3. The method for active thermal management control of engineering machinery based on deep learning as claimed in claim 1, characterized in that: The thermal management parameters of the machine include engine coolant temperature, hydraulic system coolant temperature and the heat dissipation of each cooling module; Powertrain operating parameters include engine speed, engine torque, hydraulic cylinder load, and construction machinery travel speed; For the thermal management system of construction machinery, different ambient temperatures, engine speeds, engine torques, hydraulic cylinder loads, engine coolant temperatures, hydraulic system coolant temperatures, heat dissipation of the cooling module, and construction machinery travel speeds are set as training sample inputs. The fan speeds that meet the safety temperature requirements of the construction machinery are simulated and used as training samples for the prediction model. The improved support vector machine is used to train the samples to form a prediction model for fan speed.

4. The method for active thermal management control of construction machinery based on deep learning as claimed in claim 1, characterized in that: The process of predicting the next micro-load segment using the Markov chain and association rule algorithm includes: Based on the micro-load fragment of the current load spectrum, calculate the support of all 2-item sets, the current state Transition to next state , the item set The support and confidence are: ； ； Support Represents all itemsets Likelihood of occurrence, confidence Indicates prerequisites If another transaction occurs The probability of occurrence, P represents the probability, calculated in the current state The maximum support of all item sets under , all 2-itemset supports are normalized, and the normalization formula is: ； The normalized support and confidence are used to obtain the association rule score using the following function: ； ； ； in, It is an item set The support of It is an item set The confidence level, It is an item set Association rule scoring; The improved state transition probability is obtained by multiplying the association rule score with the corresponding state transition probability, and the working condition with the maximum improved transition probability is selected as the prediction result of the next state load.

5. The method for active thermal management control of engineering machinery based on deep learning as claimed in claim 1, characterized in that: The real-time ambient temperature, engine coolant temperature, hydraulic system coolant temperature, heat dissipation of the cooling module, cooling wind speed, as well as engine speed, engine torque and hydraulic cylinder load under weighted working conditions are used as inputs of the fan speed prediction model to predict the fan speed. The fan speed is converted into the PWM duty cycle that drives the fan variable plunger pump speed to form a control strategy.

6. An active thermal management control system for construction machinery based on deep learning, characterized in that: include: A deep learning algorithm improvement module is configured to improve a support vector machine using a deep learning algorithm to obtain an improved support vector machine algorithm, and to train historical data samples including ambient temperature, thermal management parameters of construction machinery, powertrain operating parameters, and fan speed using the improved support vector machine algorithm to form a prediction model for fan speed; The specific process of using deep learning algorithm to improve support vector machine includes: introducing the whale encircling prey and spiral bubble attack prey position update strategies in the whale algorithm into the artificial rabbit algorithm, improving its detour foraging and random hiding behaviors, obtaining the artificial rabbit and whale fusion algorithm, and using the artificial rabbit and whale fusion algorithm to improve support vector machine; The specific process of improving the support vector machine by using the artificial rabbit and whale fusion algorithm includes: the support vector machine algorithm based on the fusion of the artificial rabbit and whale algorithms sets initial parameters, randomly generates a combination of the variance parameters of the penalty factor and the radial basis kernel function as the initial population, and searches for a combination of the variance parameters of the penalty factor and the radial basis kernel function that minimizes the error of the support vector machine prediction model; The micro-load segment prediction module is configured to obtain the micro-load segment of the current load spectrum, establish the probability transfer matrix of various working conditions by using the Markov chain model, calculate the support and confidence of the relevant item sets by using the association rule algorithm, and construct the association rule score based on the support and confidence, and combine it with the transfer probability of the corresponding state to predict the next micro-load segment; A weighted operating condition parameter calculation module is configured to calculate a powertrain operating parameter of the construction machinery under a weighted operating condition based on weights of a current microload segment and a predicted microload segment; Based on the weights of the current micro-load segment and the predicted micro-load segment, the specific process of calculating the powertrain operating parameters of the construction machinery under the weighted working condition includes: The objective weight is determined according to the difference between the load spectra of the two fragments. The weight of the moment is: ； in , for The load of construction machinery at all times, for The load of construction machinery at all times, for The weight of the moment, is the coefficient, is a constant; because Moment and The moment load weight correlation coefficient is -1, and the weighted load is: ； The fan speed prediction module is configured to use the real-time ambient temperature, thermal management parameters and powertrain operating parameters under weighted working conditions as input quantities of the prediction model, predict the fan speed, and convert it into a control signal.

7. A computer-readable storage medium, characterized in that: A computer program is stored thereon, which, when executed by a processor, implements the steps in the deep learning-based active thermal management control method for engineering machinery as described in any one of claims 1 to 5.

Images