Method and apparatus for recommending function configuration parameters of a working machine

Abstract

The application relates to the technical field of intelligent control of engineering machinery, and discloses a recommendation method and device for function configuration parameters of engineering machinery, which collects operation data, working environment data and operation data of an operation subject of a target device; identifies working condition data of the target device according to the operation data, the working environment data and a preset time sequence deep learning model; constructs and dynamically updates behavior portrait data for representing behavior characteristics of the operation subject according to the operation data; inputs the working condition data and the behavior portrait data into a preset hybrid decision model to output target function configuration parameters, the preset hybrid decision model is used for performing multi-target optimization operation and internally building device operation protection logic configured according to the working condition data; the target function configuration parameters are issued to the target device for execution; and the preset hybrid decision model is updated based on feedback data after the execution. It can be seen that the application can improve the recommendation accuracy of the function configuration parameters of the engineering machinery.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for construction machinery, and in particular to a method and apparatus for recommending functional configuration parameters for construction machinery. Background Technology

[0002] Modern construction machinery (such as excavators) are generally equipped with electronic control systems. Drivers can manually set function configuration parameters such as working mode, throttle gear, and operating sensitivity through a human-machine interface. These parameters directly affect the equipment's operating efficiency, fuel economy, and operating experience.

[0003] Currently, the configuration of functional parameters mainly relies on the driver's subjective experience, and the system lacks an intelligent guidance mechanism. Since the parameters need to take into account multiple dimensions such as efficiency, fuel consumption, and equipment safety, and the requirements for parameters vary significantly in different operating scenarios (such as mining and municipal construction), ordinary drivers find it difficult to accurately match actual needs. This often results in parameter settings that cannot adapt to the characteristics of the working conditions and operating habits, affecting the full utilization of equipment performance.

[0004] Therefore, it is particularly important to propose a technical solution to improve the accuracy of recommended functional configuration parameters for engineering machinery. Summary of the Invention

[0005] This invention provides a method and apparatus for recommending functional configuration parameters of construction machinery, which can improve the accuracy of recommending functional configuration parameters of construction machinery.

[0006] To address the aforementioned technical problems, the first aspect of this invention discloses a method for recommending functional configuration parameters for engineering machinery, the method comprising: Collect operational data of the target equipment, operating environment data, and operational data of the main operating entity of the target equipment; Based on the operational data, the working environment data, and the preset time-series deep learning model, the working condition data of the target device is identified, and the working condition data is used to at least indicate the working mode and working object of the target device in the current environment; Based on the operation data, construct and dynamically update behavioral profile data to characterize the behavioral features of the operation subject; The operation condition data and the behavior profile data are input into a preset hybrid decision model, and the target function configuration parameters are output. The preset hybrid decision model is used to perform multi-objective optimization calculations including operation efficiency, energy consumption level and equipment loss, and has built-in equipment operation protection logic configured according to the operation condition data. The target function configuration parameters are sent to the target device for execution. The preset hybrid decision model is updated based on the feedback data after execution.

[0007] As an optional implementation, in the first aspect of the present invention, identifying the operating condition data of the target device based on the operating data, the operating environment data, and a preset time-series deep learning model includes: The runtime data and the work environment data are spatiotemporally aligned to obtain a multidimensional feature sequence; The multidimensional feature sequence is input into a preset temporal deep learning model, and the state change features of the target device in a continuous time period are extracted through a self-attention mechanism to obtain initial working condition data. Obtain a predefined work condition knowledge graph, which is a structured data set containing the logical relationships between work methods, attributes of work objects, work environment restrictions, and equipment work restrictions of the target equipment; The initial operating condition data is mapped to the operating condition knowledge graph for matching, and the operating condition data that matches the logical relationship in the operating condition knowledge graph is filtered to determine the operating condition data of the target equipment.

[0008] As an optional implementation, in the first aspect of the present invention, the step of constructing and dynamically updating behavioral profile data to characterize the behavioral features of the operating subject based on the operational data includes: The operation data is used to extract a first quantitative feature value, which includes operation category, operation frequency, operation range and operation duration. The first quantitative feature value is used to indicate the intensity and duration of the operation subject under different work actions. Behavioral profile data is determined based on the first quantified feature value, and the behavioral profile data includes at least one feature label among operational aggressiveness, efficiency preference and proficiency. Based on the second quantized feature value corresponding to the historical operation data, the feature label in the behavior profile data is iteratively corrected, and the corrected behavior profile data is used as the current behavior profile data, wherein the second quantized feature value corresponds to the first quantized feature value.

[0009] As an optional implementation, in the first aspect of the present invention, the step of inputting the work condition data and the behavioral profile data into a preset hybrid decision model and outputting target function configuration parameters includes: The operational condition data and the behavioral profile data are fused to obtain fused decision features; Based on the fusion decision features, the operation strategy corresponding to the operation condition data under the current operation scenario is determined. The operation strategy is used to indicate the priority allocation relationship between operation efficiency, energy consumption level and equipment wear. Based on the operational strategy, multi-objective optimization is performed on at least one candidate parameter combination corresponding to the fusion decision feature to obtain a comprehensive evaluation value for each candidate parameter combination. Based on the device operation protection logic built into the preset hybrid decision model, candidate parameter combinations that do not conform to the device operation protection logic are filtered out. The device operation protection logic is used to define the allowable range of parameters for the target device in the current operating environment. The parameter with the highest comprehensive evaluation value among the remaining candidate parameter combinations after filtering is determined as the target function configuration parameter.

[0010] As an optional implementation, in the first aspect of the present invention, updating the preset hybrid decision model based on feedback data after execution includes: The feedback data generated during the execution of the target function configuration parameters by the target device is obtained. The feedback data includes the adjustment record of the target function configuration parameters by the operating subject and the actual operating deviation of the target device during the execution process. Based on the feedback data, the correction parameters of the preset hybrid decision model are determined. The correction parameters are used to adjust the priority allocation relationship in the multi-objective optimization operation and / or the allowable range of parameters in the equipment operation protection logic. The hybrid decision model is updated based on the corrected parameters.

[0011] As an optional implementation, in the first aspect of the present invention, the method further includes: The work status data and corresponding behavior profile data generated within a preset time period are statistically analyzed to construct a group work dataset, which is used to reflect the parameter configuration preferences of different operating subjects in different work scenarios. The group operation dataset is input into the preset hybrid decision model to update the historical benchmark data in the multi-objective optimization operation; When the behavior profile data of the target device is detected to deviate from the common features in the group operation dataset by more than a preset threshold, the target function configuration parameters are calibrated according to the group operation dataset, and the calibrated target function configuration parameters are used as the recommended output.

[0012] As an optional implementation, in the first aspect of the present invention, the method further includes: In response to the target device being in a standby state during a work interval, the standby time of the target device and the current remaining work tasks are obtained; Based on the standby time and the remaining task data, predict the expected operating condition change trend of the target equipment in the next operation stage; If the expected operating condition change trend indicates that an operating condition type switch is about to occur, the equipment operation protection logic that matches the expected operating condition change trend is invoked in advance, and the priority allocation relationship of the multi-objective optimization operation in the preset hybrid decision model is pre-adjusted to generate pre-adapted target function configuration parameters.

[0013] A second aspect of the present invention discloses a device for recommending functional configuration parameters of engineering machinery, the device comprising: The data acquisition module is used to collect the operating data of the target device, the operating environment data, and the operating data of the main body operating the target device; The identification module is used to identify the operating condition data of the target device based on the running data, the operating environment data and the preset time-series deep learning model. The operating condition data is used to indicate at least the operating mode and operating object of the target device in the current environment. The construction module is used to construct and dynamically update behavioral profile data that characterizes the behavioral features of the operating subject based on the operation data. The calculation module is used to input the work condition data and the behavior profile data into the preset hybrid decision model and output the target function configuration parameters. The preset hybrid decision model is used to perform multi-objective optimization calculations including work efficiency, energy consumption level and equipment loss, and has built-in equipment operation protection logic configured according to the work condition data. The execution module is used to send the target function configuration parameters to the target device for execution; The update module is used to update the preset hybrid decision model based on the feedback data after execution.

[0014] As an optional implementation, in a second aspect of the present invention, the specific method by which the identification module identifies the operating condition data of the target device based on the operating data, the operating environment data, and a preset time-series deep learning model includes: The runtime data and the work environment data are spatiotemporally aligned to obtain a multidimensional feature sequence; The multidimensional feature sequence is input into a preset temporal deep learning model, and the state change features of the target device in a continuous time period are extracted through a self-attention mechanism to obtain initial working condition data. Obtain a predefined work condition knowledge graph, which is a structured data set containing the logical relationships between work methods, attributes of work objects, work environment restrictions, and equipment work restrictions of the target equipment; The initial operating condition data is mapped to the operating condition knowledge graph for matching, and the operating condition data that matches the logical relationship in the operating condition knowledge graph is filtered to determine the operating condition data of the target equipment.

[0015] As an optional implementation, in the second aspect of the present invention, the specific method by which the construction module constructs and dynamically updates behavioral profile data characterizing the behavioral features of the operating subject based on the operation data includes: The operation data is used to extract a first quantitative feature value, which includes operation category, operation frequency, operation range and operation duration. The first quantitative feature value is used to indicate the intensity and duration of the operation subject under different work actions. Behavioral profile data is determined based on the first quantified feature value, and the behavioral profile data includes at least one feature label among operational aggressiveness, efficiency preference and proficiency. Based on the second quantized feature value corresponding to the historical operation data, the feature label in the behavior profile data is iteratively corrected, and the corrected behavior profile data is used as the current behavior profile data, wherein the second quantized feature value corresponds to the first quantized feature value.

[0016] As an optional implementation, in the second aspect of the present invention, the specific method by which the computing module inputs the work condition data and the behavior profile data into a preset hybrid decision model and outputs the target function configuration parameters includes: The operational condition data and the behavioral profile data are fused to obtain fused decision features; Based on the fusion decision features, the operation strategy corresponding to the operation condition data under the current operation scenario is determined. The operation strategy is used to indicate the priority allocation relationship between operation efficiency, energy consumption level and equipment wear. Based on the operational strategy, multi-objective optimization is performed on at least one candidate parameter combination corresponding to the fusion decision feature to obtain a comprehensive evaluation value for each candidate parameter combination. Based on the device operation protection logic built into the preset hybrid decision model, candidate parameter combinations that do not conform to the device operation protection logic are filtered out. The device operation protection logic is used to define the allowable range of parameters for the target device in the current operating environment. The parameter with the highest comprehensive evaluation value among the remaining candidate parameter combinations after filtering is determined as the target function configuration parameter.

[0017] As an optional implementation, in the second aspect of the present invention, the specific method by which the updating module updates the preset hybrid decision model based on the feedback data after execution includes: The feedback data generated during the execution of the target function configuration parameters by the target device is obtained. The feedback data includes the adjustment record of the target function configuration parameters by the operating subject and the actual operating deviation of the target device during the execution process. Based on the feedback data, the correction parameters of the preset hybrid decision model are determined. The correction parameters are used to adjust the priority allocation relationship in the multi-objective optimization operation and / or the allowable range of parameters in the equipment operation protection logic. The hybrid decision model is updated based on the corrected parameters.

[0018] As an optional implementation, in the second aspect of the present invention, the construction module is further configured to statistically analyze the work condition data and the corresponding behavioral profile data generated within a preset time period to construct a group work dataset, wherein the group work dataset is used to reflect the parameter configuration preferences of different operating subjects under different work scenarios; The update module is also used to input the group operation dataset into the preset hybrid decision model to update the historical benchmark data in the multi-objective optimization operation; The device also includes: The calibration module is used to calibrate the target function configuration parameters according to the group operation dataset when the behavior profile data of the target device deviates from the general features in the group operation dataset by more than a preset threshold, and to output the calibrated target function configuration parameters as a recommendation.

[0019] As an optional implementation, in a second aspect of the invention, the apparatus further includes: The acquisition module is used to acquire the standby time and the current remaining task data of the target device in response to the target device being in a standby state during a work interval; The prediction module is used to predict the expected operating condition change trend of the target equipment in the next operation stage based on the standby time and the remaining task data. The generation module is used to, if the expected operating condition change trend indicates that an operating condition type switch is about to occur, call in advance the equipment operation protection logic that matches the expected operating condition change trend, and pre-adjust the priority allocation relationship of the multi-objective optimization operation in the preset hybrid decision model to generate pre-adapted target function configuration parameters.

[0020] A third aspect of the present invention discloses another device for recommending functional configuration parameters of engineering machinery, the device comprising: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the method for recommending functional configuration parameters of engineering machinery disclosed in the first aspect of the present invention.

[0021] The fourth aspect of the present invention discloses a computer storage medium storing computer instructions, which, when invoked, are used to execute the method for recommending functional configuration parameters of engineering machinery disclosed in the first aspect of the present invention.

[0022] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: In this embodiment of the invention, operational data of the target device, operating environment data, and operational data of the operator of the target device are collected. Based on the operational data, operating environment data, and a preset time-series deep learning model, the operating condition data of the target device is identified. The operating condition data at least indicates the operating mode and operating object of the target device in the current environment. Based on the operational data, behavioral profile data characterizing the behavioral features of the operator is constructed and dynamically updated. The operating condition data and behavioral profile data are input into a preset hybrid decision model, which outputs target function configuration parameters. The preset hybrid decision model performs multi-objective optimization calculations including operating efficiency, energy consumption level, and equipment wear, and incorporates equipment operation protection logic configured based on the operating condition data. The target function configuration parameters are then sent to the target device for execution. The preset hybrid decision model is updated based on the feedback data after execution. Therefore, implementing this invention can improve the comprehensiveness of data dimensions by collecting operational data of the target device, operating environment data, and operational data of the operator, thereby facilitating the construction of a more complete equipment status view, which in turn improves the accuracy of subsequent decisions and enables intelligent adaptation of the equipment operation process. It can identify operational condition data through a pre-set time-series deep learning model, improving the dynamic adaptability of condition identification. This helps capture the continuous changing characteristics of equipment operation, avoiding the lag of static identification and achieving real-time and accurate condition judgment. It can perform multi-objective optimization calculations, including operational efficiency, energy consumption level, and equipment wear, through a pre-set hybrid decision model, improving the rationality of multi-objective trade-offs. This helps balance the efficiency requirements and resource consumption of equipment operation, thereby reducing the overall cost of the equipment's entire life cycle and achieving sustainable operation. It can improve the accuracy of recommended functional configuration parameters for construction machinery and enhance equipment operation safety through built-in equipment operation protection logic configured based on operational condition data. This helps avoid the risk of damage to equipment due to exceeding parameter limits, thereby extending equipment life and achieving a balance between operational safety and performance. It can update the pre-set hybrid decision model based on post-execution feedback data, improving the model's dynamic evolution capability. This helps adapt to long-term changes in the equipment's operating environment and operating habits, thereby improving the long-term effectiveness of parameter recommendations and achieving model self-optimization and continuous iteration. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart illustrating a method for recommending functional configuration parameters for engineering machinery, as disclosed in an embodiment of the present invention. Figure 2 This is a flowchart illustrating another method for recommending functional configuration parameters for engineering machinery, as disclosed in an embodiment of the present invention. Figure 3 This is a schematic diagram of the structure of a device for recommending functional configuration parameters of engineering machinery, as disclosed in an embodiment of the present invention. Figure 4 This is a schematic diagram of the structure of another device for recommending functional configuration parameters of engineering machinery disclosed in an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of another device for recommending functional configuration parameters of engineering machinery disclosed in an embodiment of the present invention. Detailed Implementation

[0025] To enable those skilled in the art to better understand the present invention, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0027] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0028] This invention discloses a method and apparatus for recommending functional configuration parameters for construction machinery. By collecting operational data from the target equipment, the working environment, and the operator, it improves the comprehensiveness of data dimensions, thereby facilitating the construction of a more complete equipment status view. This, in turn, improves the accuracy of subsequent decision-making and enables intelligent adaptation of the equipment operation process. It can identify operational condition data through a preset temporal deep learning model, improving the dynamic adaptability of condition identification. This helps capture the continuous changing characteristics of equipment operation, avoiding the lag of static identification and achieving real-time and accurate condition judgment. It can perform multi-objective optimization calculations, including operational efficiency, energy consumption level, and equipment wear, through a preset hybrid decision model, improving the rationality of multi-objective trade-offs. This helps balance the efficiency requirements and resource consumption of equipment operation, thereby reducing the overall cost of the equipment's entire life cycle and achieving sustainable operation. By incorporating equipment operation protection logic configured based on operational condition data, it improves the accuracy of recommending functional configuration parameters for construction machinery and enhances equipment operation safety. This helps avoid the risk of damage to equipment due to exceeding parameters, thereby extending equipment life and achieving a balance between operational safety and performance. It can update the preset hybrid decision-making model based on feedback data after execution, thereby improving the model's dynamic evolution capability. This facilitates adaptation to long-term changes in equipment operating environment and operating habits, and further enhances the long-term effectiveness of parameter recommendations, enabling the model to self-optimize and continuously iterate. These will be explained in detail below.

[0029] Example 1 Please see Figure 1 , Figure 1 This is a flowchart illustrating a method for recommending functional configuration parameters of engineering machinery, as disclosed in an embodiment of the present invention. Figure 1 The recommended method for functional configuration parameters of construction machinery described herein can be applied to construction machinery, such as excavators, and can also be applied to intelligent devices associated with construction machinery. These intelligent devices include, but are not limited to, one or more of the following: switching devices, cloud devices, edge computing devices, relay devices, base station devices, urban management devices, and intelligent connected devices. This invention does not limit the scope of these applications. Figure 1 As shown, the recommended method for configuring the functional parameters of this construction machinery may include the following operations: 101. Collect the target equipment's operational data, operating environment data, and the operational data of the target equipment's operating entity; In this embodiment of the invention, optionally, the following data can be collected: operating data of the target equipment, operating environment data, and operating data of the target equipment's operator. Operating data specifically refers to the physical and state quantities generated during equipment operation, including engine speed, hydraulic system pressure, overall load rate, instantaneous fuel consumption, and temperature information of core components. Operating environment data refers to sensor data of the external environment in which the equipment is located, including geographical coordinates, ambient temperature, air humidity, and altitude. Operating data of the operator refers to the input signals from the driver on the control panel, including changes in the position of the joystick, pedal opening, and the currently selected operating mode.

[0030] 102. Based on the operating data, working environment data and the preset time-series deep learning model, identify the working condition data of the target equipment. The working condition data shall at least indicate the working mode and working object of the target equipment in the current environment. Optionally, in this embodiment of the invention, the operating condition data of the target equipment is identified based on operational data, operating environment data, and a preset time-series deep learning model. The operating condition data can be used to describe the current operating state of the equipment, and at least includes the operating method (such as excavation, loading, leveling) and the operating object (such as earthwork, rock, and gravel).

[0031] In this embodiment of the invention, as an optional implementation, the above-mentioned method of identifying the operating condition data of the target device based on operating data, operating environment data, and a preset time-series deep learning model includes: Spatiotemporal alignment of operational data and work environment data yields a multidimensional feature sequence. The multidimensional feature sequence is input into the preset temporal deep learning model, and the state change features of the target device in a continuous time period are extracted through the self-attention mechanism to obtain the initial working condition data. Obtain a predefined work condition knowledge graph, which is a structured data set containing the logical relationships between work methods, work object attributes, work environment constraints, and target equipment work constraints; The initial operating condition data is mapped to the operating condition knowledge graph for matching, and the operating condition data that matches the logical relationship in the operating condition knowledge graph is filtered to determine the operating condition data of the target equipment.

[0032] Optionally, in this embodiment of the invention, the operating condition data of the target device can be identified based on the operating data, the working environment data, and the preset time-series deep learning model.

[0033] Spatiotemporal alignment of operational data and working environment data yields a multidimensional feature sequence. Spatiotemporal alignment refers to arranging various types of data collected at different time points according to a unified timestamp, eliminating data misalignment caused by different sampling frequencies, and forming a multidimensional data matrix containing the time dimension.

[0034] The multidimensional feature sequence is input into a pre-defined temporal deep learning model, and the state change features of the target equipment over a continuous time period are extracted through a self-attention mechanism to obtain initial operating condition data. The self-attention mechanism can be used to calculate the correlation weights between data points at different times in the sequence, capture the evolution pattern of equipment actions, such as identifying the continuous action feature of "changing from high-speed walking to low-speed high-torque digging".

[0035] Obtain a predefined operating condition knowledge graph. An operating condition knowledge graph is a structured data set containing logical relationships between operating methods, attributes of operating objects, operating environment constraints, and equipment operating constraints of the target equipment. For example, the graph defines the "rock" attribute as corresponding to the "high equipment wear" logic, and the "plateau environment" attribute as corresponding to the "engine power limited" logic.

[0036] The initial operating condition data is mapped to an operating condition knowledge graph for matching. Operating condition data that matches the logical relationships in the knowledge graph is then filtered to determine the operating condition data of the target equipment. This step is used to verify whether the model's identification results conform to common sense in physics; for example, logically contradictory identification results such as "crushing operations in soft soil" are eliminated, ensuring that the final determined operating condition data is logically reasonable.

[0037] In this embodiment of the invention, optionally, a hybrid architecture combining an improved Gated Recurrent Unit (GRU) and a self-attention mechanism can be adopted for the preset temporal deep learning model. The model input layer receives a multi-dimensional feature sequence (e.g., N×M, where N is the time step and M is the feature dimension), captures short-term dependent features through two GRU layers, and then inputs the GRU output to a multi-head self-attention layer (e.g., 4 heads) to calculate the global association weights of features at different time points. Supervised learning is used during the model training phase. The training dataset includes labeled work condition data (labeled as work method and work object), the loss function is a weighted combination of cross-entropy loss and KL divergence, the optimizer is AdamW, and the learning rate is set to 0.001 and decays with each training epoch. Through this training process, the model can accurately identify the temporal evolution patterns of continuous actions such as "digging-loading-leveling," rather than a general classification model.

[0038] In this embodiment of the invention, optionally, a sliding window mechanism (e.g., a window length of 10 consecutive sampling periods) can be used for the input of multidimensional feature sequences. Each time data from a new time point is input, the window slides forward one step, updating the initial working condition data in real time. The matching process of the working condition knowledge graph supports fuzzy matching: when the similarity between the initial working condition data and a certain logical relationship in the graph exceeds, for example, 85%, it is determined to be the working condition category; if multiple categories with similarity exceeding, for example, 85% exist, the one with the highest similarity is taken as the final result. Furthermore, the graph supports dynamic updates: each month, logical relationships are automatically added or corrected based on newly collected working condition data (e.g., adding an association between "frozen soil working condition" and "preheating logic").

[0039] Further optionally, various sliding window strategies can be employed for the input of multi-dimensional feature sequences: such as fixed-length windows, adaptive-length windows (dynamically adjusted according to data fluctuations), or overlapping windows (adjacent windows share some data). The matching process of the work condition knowledge graph supports multiple logical verifications: such as fuzzy matching based on semantic similarity, precise matching based on rule reasoning, or multi-level matching (matching the work method first, then matching the work object). The dynamic update mechanism of the graph may include: regular manual maintenance, automatic incremental updates based on new data, or sharing updates with other devices through federated learning.

[0040] Further optional, the implementation steps for the adaptive length window are as follows: calculate the rate of change of data at adjacent time points; when the rate of change exceeds a preset fluctuation threshold, shorten the window length to capture rapid changes; when the rate of change is lower than the preset fluctuation threshold, extend the window length to stabilize feature extraction.

[0041] The implementation steps for multi-level matching are as follows: First, extract the keywords of the work mode from the initial working condition data and perform a first-level match with the work mode nodes in the graph; if the first-level match is successful, extract the keywords of the work object and perform a second-level match with the work object nodes of the corresponding work mode in the graph; if the second-level match is successful, the final working condition is determined.

[0042] The implementation steps of the Transformer architecture are as follows: input the multidimensional feature sequence into the embedding layer of the Transformer encoder, calculate the global dependency weights of each time step through the self-attention mechanism, and output a high-dimensional feature vector as the initial working condition data through the feedforward neural network.

[0043] The implementation steps for pattern matching based on a rule engine are as follows: convert the initial working condition data into factual input for the rule engine, load the logical rules in the graph (such as "rock working condition → high wear limit"), the rule engine matches the facts with the rules, and outputs working condition data that conforms to the logic.

[0044] The implementation steps for integrating the expert experience rule base are as follows: collect the parameter configuration experience of industry experts for specific working conditions, organize it into rule entries of "operation mode + operation object → parameter range", and import it into the graph as logical constraint nodes.

[0045] The verification steps for combining engine temperature are as follows: obtain the current engine temperature; if the temperature exceeds the preset high temperature threshold and the identified operating condition is high load operation, the operating condition identification result is determined to be valid; otherwise, the operating condition identification process is retried.

[0046] Alternatively, the initial operating condition data can be generated based on various feature extraction architectures: such as convolutional neural networks for spatial feature extraction, long short-term memory networks for long temporal dependency capture, or Transformer architecture for global feature modeling. The matching of the operating condition knowledge graph can be combined with various reasoning mechanisms: such as ontology-based logical reasoning, graph neural network-based embedding reasoning, or rule engine-based pattern matching.

[0047] As can be seen, implementing this optional embodiment can improve the temporal consistency of multi-source data by aligning operational data and work environment data in time and space, thereby helping to eliminate time deviations in data collection and facilitating the construction of coherent multi-dimensional feature sequences, thus ensuring the reliability of the data foundation. It can extract state change features within continuous time periods through the self-attention mechanism of a pre-defined temporal deep learning model, improving the depth and correlation of feature extraction, thereby helping to capture the evolutionary patterns of equipment actions and enhancing the representational ability of initial working condition data, enabling accurate depiction of complex work scenarios. It can match and filter working condition data through a predefined working condition knowledge graph, improving the logical rigor of working condition identification, thus helping to eliminate misidentification results that do not conform to physical common sense, ensuring the consistency between operational working condition data and real working scenarios, and ensuring the rationality of working condition judgments. It can improve the correlation between working condition data and work methods and work object attributes through the structured logical relationships of the working condition knowledge graph, thereby helping to clarify the parameter configuration boundaries under different work scenarios, providing a reliable basis for subsequent decision-making, and achieving scenario adaptability of parameter recommendations.

[0048] 103. Based on the operational data, construct and dynamically update behavioral profile data to characterize the behavioral features of the operational subject; In this embodiment of the invention, optionally, behavioral profile data is constructed and dynamically updated based on operational data to characterize the behavioral features of the operator. The behavioral profile data can be a digital abstraction of the driver's operating habits, reflecting whether the driver tends towards aggressive or conservative operation, and their preference for operational efficiency or fuel economy.

[0049] In this embodiment of the invention, as another optional implementation, the above-mentioned construction and dynamic updating of behavioral profile data to characterize the behavioral features of the operating subject based on operational data includes: The first quantitative feature value is extracted from the operation data, which includes operation category, operation frequency, operation range and operation duration. The first quantitative feature value is used to indicate the intensity and duration of the operation subject under different operation actions. Behavioral profile data is determined based on the first quantitative feature value. The behavioral profile data includes at least one feature label among operational aggressiveness, efficiency preference, and proficiency. Based on the second quantitative feature value corresponding to the historical operation data, the feature labels in the behavior profile data are iteratively corrected, and the corrected behavior profile data is used as the current behavior profile data. The second quantitative feature value corresponds to the first quantitative feature value.

[0050] In this embodiment of the invention, optionally, behavioral profile data for characterizing the behavioral features of the operating subject is constructed and dynamically updated based on the operation data.

[0051] The first quantitative feature value is extracted from the operation data, including operation category, operation frequency, operation amplitude, and operation duration. This first quantitative feature value indicates the intensity and duration of the operator's actions under different operational conditions. Operation category refers to the specific type of action, such as boom lifting or stick retraction; operation frequency refers to the number of times the action occurs per unit time; operation amplitude refers to the distance the control stick deviates from its neutral position; and operation duration refers to the duration of a single action.

[0052] Behavioral profile data is determined based on the first quantitative feature value. The behavioral profile data includes at least one feature label from operational aggression, efficiency preference, and proficiency. Operational aggression is determined by the feature value of high throttle and fast movements; efficiency preference is determined by the feature value of the number of operation cycles per unit time; proficiency is determined by the proportion of compound movements (simultaneous operation of two or more handles) and the smoothness of movements.

[0053] Based on the second quantified feature value corresponding to historical operation data, the feature labels in the behavioral profile data are iteratively corrected. The second quantified feature value corresponds to the first quantified feature value and refers to historical data statistically collected over a past period. The iterative correction adopts a smooth update mechanism, that is, the currently calculated feature value is merged with the historical feature value in a certain proportion to prevent the profile data from changing abruptly due to a single abnormal operation. The corrected behavioral profile data is then used as the current behavioral profile data.

[0054] In this embodiment of the invention, optionally, the extraction of the first quantitative feature value can also be combined with the temporal correlation of the operation data: for example, the operation frequency can be calculated as the number of times the joystick moves from the center to the maximum amplitude per unit time, and the operation amplitude can be calculated as the ratio of the average distance of the joystick deviating from the center to the maximum distance. The feature labels of the behavioral profile data support dynamic adjustment: if the quantitative feature value corresponding to the operation aggression label fluctuates by more than 20% within three consecutive operation cycles, the weight of that label is automatically reduced by 50% to avoid a single label excessively influencing the decision.

[0055] Further, the extraction of the first quantified feature value can employ various computational logics: such as statistical features (mean, variance), time-series features (rate of change, acceleration), or frequency-domain features (spectral energy). The feature labels for behavioral profile data support multiple dynamic adjustment strategies: such as time-decay-based weight adjustment (higher weight for recent data), confidence-based weight adjustment (higher weight for higher data quality), or scenario-based adaptive adjustment (different label weights under different working conditions).

[0056] Furthermore, the behavioral profile data can be stored using various structures: such as vectorized representation, key-value pair storage, or graph structure storage (reflecting the relationships between tags). The profile update mechanism can support multiple triggering conditions: such as time-triggered (daily updates), event-triggered (updates when operating modes change), or performance-triggered (updates when feedback data deviation exceeds a threshold).

[0057] Further optional, the implementation steps for the time-series-based rate of change are as follows: calculate the maximum difference in the change of the joystick position per unit time as the rate of change of the operation amplitude, and statistically analyze the variance of the fluctuation of the operation frequency per unit time as an operation stability index.

[0058] The implementation steps for weight adjustment based on time decay are as follows: set a time decay factor, the feature value corresponding to the operation data that is closer to the current time has a higher weight, and the weight of historical data decreases exponentially over time.

[0059] The implementation steps for vectorized representation are as follows: map labels such as operational aggressiveness, efficiency preference, and proficiency into high-dimensional vectors, train the vector representation of each label through the embedding layer, and store it as the feature vector of the behavior profile.

[0060] The implementation steps for event triggering are as follows: monitor the operation mode switching event of the operating subject (such as switching from manual mode to automatic mode), and trigger the recalculation and update of the behavior profile data.

[0061] The steps for implementing abnormal operation warnings are as follows: compare real-time operation data with normal operation feature vectors in the behavior profile; if the deviation exceeds a preset threshold, an abnormal operation warning signal is generated.

[0062] The steps for implementing eigenvalue normalization are as follows: map the quantized eigenvalues ​​such as operation frequency and amplitude to the [0,1] interval, eliminate the sensitivity of the original data, and store the normalized eigenvalues.

[0063] As can be seen, implementing this optional embodiment can improve the objectivity of behavioral feature quantification by extracting first quantitative feature values ​​of operation category, operation frequency, operation magnitude, and operation duration from operation data. This helps avoid the bias of subjective evaluation and, consequently, accurately reflects the intensity and duration of the operator's actions, thus establishing a data-driven foundation for behavioral profiling. It can determine behavioral profile data containing operational aggression, efficiency preferences, and proficiency based on the first quantitative feature values, improving the structured expression of operational habits. This helps transform implicit operational experience into explicit labels, providing personalized input features for decision-making models and enabling customized parameter recommendations. It can iteratively correct behavioral profile labels using second quantitative feature values ​​corresponding to historical operation data, improving the dynamic adaptability of behavioral profiles. This helps eliminate the impact of single abnormal operations, reflecting the long-term stable characteristics of the operator and ensuring the authenticity and reliability of profile data. A smooth update mechanism can fuse current and historical feature values, improving the stability of profile data. This helps avoid drastic changes in the profile due to short-term operational fluctuations, ensuring the consistency of parameter recommendations and achieving stable user experience.

[0064] 104. Input the work condition data and behavior profile data into the preset hybrid decision model, and output the target function configuration parameters. The preset hybrid decision model is used to perform multi-objective optimization calculations including work efficiency, energy consumption level and equipment loss, and has built-in equipment operation protection logic configured according to the work condition data. Optionally, in this embodiment of the invention, the target function configuration parameters are output after inputting operational condition data and behavioral profile data into a preset hybrid decision model. The preset hybrid decision model can be used to perform multi-objective optimization calculations including operational efficiency, energy consumption level, and equipment wear, i.e., finding the optimal balance point among multiple mutually constraining objectives. The model also incorporates equipment operation protection logic configured based on operational condition data to prevent the output parameters from exceeding the physical limits of the equipment. The target function configuration parameters specifically refer to the specific instruction values ​​issued to the controller for execution, such as throttle gear values ​​and hydraulic flow distribution ratios.

[0065] Optionally, the pre-defined hybrid decision-making model employs a core architecture combining deep reinforcement learning and constraint-satisfaction planning. The decision network uses an actor-critic architecture: the actor network outputs the probability distribution of working modes, throttle gear positions, and personalized operating parameters based on current scene features and driver profiles; the critic network evaluates the long-term value of state-action pairs and optimizes the actor network's strategy output through temporal difference errors. The reward function is designed as a multi-objective weighted form, comprising four components: operational efficiency, fuel consumption, equipment wear and tear, and driver satisfaction. The weight coefficients are dynamically adjusted according to operational strategies (e.g., increasing the weight of operational efficiency when rushing to meet deadlines). Safety constraints are hard-coded into insurmountable boundary conditions by the constraint satisfaction solver, ensuring the safety of recommended parameters under different environments.

[0066] Optionally, in addition to operating mode and throttle gear, the target function configuration parameters also include personalized operation parameters, such as motion sensitivity, response characteristics, and priority strategy. The recommendation logic for personalized parameters is as follows: adjust motion sensitivity based on the aggressiveness tag in the behavioral profile (high sensitivity for aggressive drivers, low sensitivity for conservative drivers); adjust response characteristics based on efficiency preference tags (shorten response latency when efficiency is prioritized, extend response buffer when energy saving is prioritized); and adjust priority strategy based on proficiency tags (experienced drivers have more customization permissions, while novices are restricted to high-risk parameters).

[0067] In another optional implementation of this invention, the above-mentioned input of work condition data and behavioral profile data into a preset hybrid decision model, and the output of target function configuration parameters, includes: Feature fusion is performed on operational condition data and behavioral profile data to obtain fused decision features; Based on the fusion decision characteristics, the operation strategy corresponding to the current operation scenario is determined by the operation condition data. The operation strategy is used to indicate the priority allocation relationship between operation efficiency, energy consumption level and equipment wear. Based on the operational strategy, multi-objective optimization is performed on at least one candidate parameter combination corresponding to the fusion decision features to obtain the comprehensive evaluation value of each candidate parameter combination; Based on the built-in equipment operation protection logic of the preset hybrid decision model, candidate parameter combinations that do not conform to the equipment operation protection logic are filtered out. The equipment operation protection logic is used to define the allowable range of parameters for the target equipment in the current operating environment. The parameter with the highest overall evaluation value among the remaining candidate parameter combinations after filtering is selected as the target function configuration parameter.

[0068] In this embodiment of the invention, optionally, the work condition data and behavior profile data are input into a preset hybrid decision model, and the target function configuration parameters are output.

[0069] Feature fusion is performed on operational condition data and behavioral profile data to obtain fused decision features. Feature fusion refers to mapping operational condition data describing the environment and behavioral profile data describing people onto the same feature space and splicing them together to form a comprehensive feature vector containing joint "scene-person" information.

[0070] Based on the fusion decision characteristics, the operational strategy corresponding to the current operational scenario is determined from the operational condition data. The operational strategy is used to indicate the priority allocation relationship between operational efficiency, energy consumption level, and equipment wear and tear. For example, in a scenario where deadlines are tight, the operational strategy is set to "efficiency first"; in a long-distance transportation scenario, it is set to "energy consumption first".

[0071] Based on the operational strategy, multi-objective optimization is performed on at least one candidate parameter combination corresponding to the fusion decision characteristics to obtain a comprehensive evaluation value for each candidate parameter combination. A candidate parameter combination refers to the set of all available parameter configuration schemes. The multi-objective optimization operation scores each candidate scheme in terms of efficiency, energy consumption, and loss according to the priority set by the operational strategy, and the scores are weighted and summed to obtain the comprehensive evaluation value.

[0072] Based on the built-in equipment operation protection logic of the pre-defined hybrid decision model, candidate parameter combinations that do not conform to the equipment operation protection logic are filtered out. The equipment operation protection logic is used to define the allowable range of parameters for the target equipment in the current operating environment. For example, in operating conditions at an altitude of over 4000 meters, the logic stipulates that the throttle limit should not exceed level 8. The filtering process eliminates parameter combinations that exceed this range.

[0073] The parameter with the highest overall evaluation value among the remaining candidate parameter combinations after filtering is selected as the target function configuration parameter.

[0074] In this embodiment of the invention, optionally, the preset hybrid decision-making model can be composed of a "scenario-profile analysis module," a "multi-objective optimization engine," and a "safety constraint verification module" connected in series. The scenario-profile analysis module maps the fused decision features to operational strategy vectors (e.g., with a dimension of 3, corresponding to the priority weights of efficiency, energy consumption, and loss, respectively); the multi-objective optimization engine uses an improved non-dominated sorting genetic algorithm to solve the Pareto front for candidate parameter combinations (e.g., a total of 500 groups, covering the feasible domain of throttle gear, hydraulic flow, and shifting logic), calculating the fitness value of each combination in the three dimensions of efficiency (e.g., workload per unit time), energy consumption (e.g., fuel consumption per 100 kilometers), and loss (e.g., temperature rise rate of key components); the safety constraint verification module has built-in equipment operation protection logic (e.g., stored in a rule base, containing multiple working condition-parameter mapping rules), filtering the Pareto front solutions and eliminating solutions that violate the rules (e.g., "throttle gear < 6 under rock conditions"). Through this architecture, the model can output optimal parameters that balance efficiency and safety under specific working conditions, rather than a general optimization tool.

[0075] In this embodiment of the invention, further optionally, various algorithm variants can be used for multi-objective optimization operations: in addition to the improved NSGA-III, a decomposition-based multi-objective evolutionary algorithm can be used to decompose the three-dimensional objectives (efficiency, energy consumption, and loss) into multiple sub-problems for parallel solution; or a reinforcement learning method can be used, with the operational strategy as the state, candidate parameter combinations as actions, and the comprehensive evaluation value as the reward, and the optimal strategy can be obtained through Q-learning training. The filtering logic of the safety constraint verification module supports hierarchy: the first level filters parameters that violate physical limits (such as engine speed > 2500 rpm), the second level filters parameters that violate operating condition limits (such as hydraulic flow < 300 L / min under rock conditions), and the third level filters parameters that violate the operating profile (such as parameters corresponding to aggressive drivers requiring a 10% safety margin).

[0076] Furthermore, multi-objective optimization operations can employ various algorithmic paradigms: such as evolutionary algorithms (genetic algorithms, particle swarm optimization), mathematical programming-based algorithms (linear weighting, ideal point method), or machine learning-based algorithms (reinforcement learning, Bayesian optimization). The filtering logic of the safety constraint verification module supports multiple hierarchical structures: such as single-layer filtering (directly removing parameters exceeding limits), multi-layer filtering (sequentially filtering physical limits, operating condition restrictions, and operational profiles), or hierarchical filtering (processing according to the severity of violations). Feature fusion methods can also be diversified: such as early fusion (directly concatenating original features), mid-stage fusion (concatenating mid-level representations), or late-stage fusion (fusion of decision results).

[0077] Further, the generation of candidate parameter combinations can employ various strategies: such as rule-based full enumeration, sampling-based random generation, or distribution fitting based on historical data. The calculation of the comprehensive evaluation value can incorporate various weighting mechanisms: such as static weighting (fixed weights), dynamic weighting (adjusted according to real-time operating conditions), or adaptive weighting (learning based on feedback data). Safety constraint verification can combine various verification logics: such as hard constraints (direct elimination), soft constraints (penalty scoring), or probabilistic constraints (based on reliability thresholds).

[0078] Alternatively, the implementation steps for rule-based full enumeration are as follows: traverse all feasible parameter combinations and select combinations that meet the constraints of the job condition data as candidate parameter combinations.

[0079] The steps for implementing static weighting are as follows: based on the operational strategy, fixed weights are preset for efficiency, energy consumption, and loss. The scores of each dimension are multiplied by their corresponding weights and then summed to obtain the comprehensive evaluation value.

[0080] The implementation steps for hard constraints are as follows: each candidate parameter combination is verified, and if the parameter value exceeds the range specified by the equipment operation protection logic, the combination is directly eliminated.

[0081] The steps for implementing multi-dimensional indicators are as follows: calculate the workload of candidate parameter combinations per unit time as an efficiency indicator, calculate the fuel consumption per unit time as an energy consumption indicator, and calculate the temperature rise rate of key components as a loss indicator.

[0082] Optionally, the equipment operation protection logic includes hard-coded constraints for multiple scenarios: throttle gears above the eighth level are not recommended in high-altitude environments to prevent engine overload; high-power mode is not recommended in high-temperature environments to reduce heat generation; throttle gears above the eighth level are not recommended in crushing operations to reduce the risk of equipment damage; and the upper limit of hydraulic flow is limited during heavy-load operations to protect the hydraulic system.

[0083] As can be seen, implementing this optional embodiment can improve the comprehensiveness of decision input by fusing features from operational condition data and behavioral profile data. This facilitates the integration of scenario features and user features, thereby enabling the construction of a joint "scenario-human" decision view and achieving two-dimensional adaptation of parameter recommendations. It can determine the operational strategy under the current operational scenario by fusing decision features, improving the dynamic adjustment capability of multi-objective priorities. This allows for flexible switching of the emphasis on efficiency, energy consumption, and loss according to operational needs, thus improving the contextual adaptability of parameter configuration and achieving optimal resource allocation. It can improve the comprehensiveness of the overall evaluation value by performing multi-objective optimization calculations on candidate parameter combinations based on operational strategies. This facilitates the evaluation of the overall performance of parameter combinations from multiple dimensions, enabling the selection of balanced solutions that take into account efficiency, energy consumption, and loss, and achieving collaborative optimization of multiple objectives. It can filter out candidate parameter combinations that do not meet the requirements through equipment operation protection logic, improving the safety of parameter output. This helps to eliminate invalid solutions that exceed physical limits or scenario limitations, ensuring that recommended parameters are within the safe operating range of the equipment and achieving controllable risks in the operational process.

[0084] 105. Send the target function configuration parameters to the target device for execution; In this embodiment of the invention, optionally, the target function configuration parameters are sent to the target device for execution. The sending process can be completed through the device's communication bus, writing the parameters into the corresponding electronic control unit.

[0085] Alternatively, before distribution, interpretability information of the recommended parameters can be displayed through a human-computer interaction interface: Scene recognition results: Clearly define the current task type, geological conditions, environmental constraints, and driver style matching degree; Reason for recommendation: Explain the basis for parameter selection (e.g., "Based on rock conditions and an aggressive driving style, heavy load mode + 7th gear throttle is recommended"). Effect prediction: Based on historical data of similar operating conditions, it shows the expected improvement in operating efficiency, reduction in fuel consumption, and equipment load level; Operation method: Supports one-click application of all recommended parameters, manual adjustment of some parameters before application or temporary non-application, and provides a memory function to reduce the burden of repeated confirmation.

[0086] 106. Update the preset hybrid decision model based on the feedback data after execution.

[0087] In this embodiment of the invention, optionally, the preset hybrid decision-making model is updated based on the feedback data after execution. Feedback data refers to the actual effect data generated after the parameters are executed. This data is used to correct the internal computational logic of the hybrid decision-making model, making it more aligned with actual needs in the next decision-making process.

[0088] Alternatively, the feedback mechanism can include a three-layer closed loop: Real-time feedback: When the driver adjusts the recommended parameters, the system records the direction and magnitude of the adjustment as a negative signal that the current recommendation does not match, which is used to correct subsequent recommendation strategies in real time; Short-term feedback: After each shift, the system calculates the deviation between the actual work efficiency and fuel consumption and the predicted values, evaluates the accuracy of the recommended strategy, and uses it to optimize the parameter output for the next shift. Long-term feedback: Monthly personalized operation reports for drivers are generated, summarizing indicators such as recommendation acceptance rate, efficiency improvement trend, and fuel consumption optimization effect. These are used to iteratively optimize the driver profile model and reward function weights, and to establish a database of excellent driver operations.

[0089] In an optional embodiment, the above-described updating of the preset hybrid decision model based on post-execution feedback data includes: Acquire feedback data generated during the execution of target function configuration parameters by the target device. The feedback data includes the operator's adjustment record of the target function configuration parameters and the actual operating deviation of the target device during the execution process. Based on the feedback data, determine the correction parameters of the preset hybrid decision model. The correction parameters are used to adjust the priority allocation relationship in the multi-objective optimization operation and / or the allowable range of parameters in the equipment operation protection logic. The hybrid decision model is updated based on the corrected parameters.

[0090] In this embodiment of the invention, optionally, the preset hybrid decision model is updated based on the feedback data after execution.

[0091] Acquire feedback data generated during the execution of target function configuration parameters by the target device. This feedback data includes records of adjustments made by the operator to the target function configuration parameters and actual operational deviations of the target device during execution. Adjustment records refer to the differences before and after the driver manually modifies the recommended parameters; if the driver frequently reduces the throttle, it indicates that the recommended values ​​are too high. Actual operational deviations refer to the differences between actual fuel consumption and efficiency and the predicted values.

[0092] Based on the feedback data, the correction parameters for the preset hybrid decision-making model are determined. These correction parameters are used to adjust the priority allocation relationships in multi-objective optimization operations and / or the allowable range of parameters in equipment operation protection logic. For example, if feedback indicates excessive fuel consumption, the correction parameters are used to reduce the weight of energy consumption indicators in multi-objective optimization operations or tighten the power limit in equipment operation protection logic.

[0093] Based on the corrected parameters, the hybrid decision model is updated. The corrected parameters are then fed back into the model training process to adjust the internal weight coefficients of the model, enabling the model to output parameters that better reflect actual feedback in subsequent decisions.

[0094] In this embodiment of the invention, optionally, the above-mentioned model update can adopt an incremental learning mechanism. After each acquisition of feedback data, only the weight matrix of the multi-objective optimization engine in the hybrid decision model is locally updated (e.g., the update ratio is 5% of the total parameters) to avoid forgetting historical knowledge caused by full training. The specific process is as follows: the feedback data (e.g., adjustment records, running deviations) is converted into training samples (the input is the fused decision features, and the label is the corrected parameter values), and mini-batch gradient descent (e.g., batch size=32) is used to train for 10 epochs, with the learning rate fixed at 0.0001. The updated model parameters are fused with the original parameters through a moving average (e.g., coefficient 0.9) to ensure model stability.

[0095] In this embodiment of the invention, optionally, the incremental learning mechanism can also be combined with transfer learning: when the target device switches to a completely new operating condition (e.g., from a plain to a plateau), the pre-trained plateau operating condition model parameters are invoked, and only a small number of iterations (e.g., 5 epochs) are performed on the feedback data under the new operating condition to quickly adapt to the new scenario. The determination of the correction parameters can also incorporate Bayesian optimization: using the feedback data as input and the correction parameters as output, the mapping relationship between the parameters and the feedback is modeled through a Gaussian process to efficiently search for the optimal correction value and reduce the number of training iterations.

[0096] Further, the model update mechanism can take several forms: such as full update (retraining the entire model), incremental update (updating only some parameters), or transfer update (reusing parameters from other models). The logic for determining the corrected parameters can include: backpropagation adjustment based on feedback data, policy gradient adjustment based on reinforcement learning, or rule-based empirical adjustment. The update frequency can also be flexibly configured: such as real-time update, periodic update, or event-triggered update (such as update when operating conditions change).

[0097] Further, the feedback data can be utilized in various forms: such as directly for gradient descent, for adjusting Bayesian priors, or for updating the rule base. Model updates can support multiple deployment strategies: such as hot updates (without service interruption), cold updates (effective after restart), or canary updates (first verified on some devices).

[0098] Further optional, the implementation steps for statistical anomaly detection are as follows: calculate the mean and standard deviation of the feedback data, mark data that deviate from the mean by more than 3 times the standard deviation as anomalies, remove the anomaly data, and then update the model.

[0099] The steps for implementing offline testing are as follows: use historical datasets to simulate the parameter recommendation process, compare the recommendation accuracy before and after the update, and if the accuracy improves, the update is confirmed to be effective.

[0100] As can be seen, implementing this optional embodiment can improve the comprehensiveness of feedback information by acquiring feedback data including operational adjustment records and actual equipment operating deviations. This facilitates the capture of deviations between the actual effects of parameter execution and the expected goals, thereby improving the reliability of model correction and achieving precise closed-loop control. The feedback data can be used to determine correction parameters, improving the targeted direction of model adjustments. This allows for optimization of weak links in priority allocation relationships or equipment operation protection logic, thereby improving the model's adaptability to actual operating scenarios and enabling continuous optimization of parameter recommendations. Furthermore, updating the hybrid decision model based on correction parameters enhances the model's self-evolution capability, facilitating the transformation of practical operating experience into the model's intrinsic knowledge and reducing long-term deviations between the model and actual needs.

[0101] As can be seen, implementing the embodiments of the present invention can improve the comprehensiveness of data dimensions by collecting the operating data of the target equipment, the operating environment data, and the operating data of the operator, thereby facilitating the construction of a more complete equipment status view, which in turn improves the accuracy of subsequent decisions and enables intelligent adaptation of the equipment operation process. It can identify operating condition data through a preset temporal deep learning model, improving the dynamic adaptability of operating condition identification, thus helping to capture the continuous changing characteristics of equipment operation and avoiding the lag of static identification, achieving real-time and accurate operating condition judgment. It can perform multi-objective optimization calculations including operating efficiency, energy consumption level, and equipment wear through a preset hybrid decision model, improving the rationality of multi-objective trade-offs, thereby balancing the efficiency requirements and resource consumption of equipment operation, and thus reducing the overall cost of the equipment's entire life cycle, achieving sustainable operation. It can improve the accuracy of recommending functional configuration parameters for construction machinery and enhance the safety of equipment operation by incorporating equipment operation protection logic configured based on operating condition data, thereby helping to avoid the risk of damage to equipment due to exceeding parameters, and thus extending the equipment's service life, achieving a balance between operational safety and performance. It can update the preset hybrid decision model based on the feedback data after execution, thereby improving the model's dynamic evolution capability. This is conducive to adapting to long-term changes in equipment operating environment and operating habits, and thus improving the long-term effectiveness of parameter recommendations, realizing the model's self-optimization and continuous iteration.

[0102] Example 2 Please see Figure 2 , Figure 2 This is a flowchart illustrating another method for recommending functional configuration parameters of engineering machinery, as disclosed in an embodiment of the present invention. Figure 2 The recommended method for functional configuration parameters of construction machinery described herein can be applied to construction machinery, such as excavators, and can also be applied to intelligent devices associated with construction machinery. These intelligent devices include, but are not limited to, one or more of the following: switching devices, cloud devices, edge computing devices, relay devices, base station devices, urban management devices, and intelligent connected devices. This invention does not limit the scope of these applications. Figure 2 As shown, the recommended method for configuring the functional parameters of this construction machinery may include the following operations: 201. Collect the target equipment's operational data, operating environment data, and the operational data of the target equipment's operating entity; 202. Based on the operational data, working environment data, and the preset time-series deep learning model, identify the working condition data of the target equipment. The working condition data shall at least indicate the working mode and working object of the target equipment in the current environment. 203. Based on the operational data, construct and dynamically update behavioral profile data to characterize the behavioral features of the operational subject; 204. Input the work condition data and behavior profile data into the preset hybrid decision model, and output the target function configuration parameters. The preset hybrid decision model is used to perform multi-objective optimization calculations including work efficiency, energy consumption level and equipment loss, and has built-in equipment operation protection logic configured according to the work condition data. 205. Send the target function configuration parameters to the target device for execution; 206. Update the preset hybrid decision model based on the feedback data after execution.

[0103] In this embodiment of the invention, it should be noted that for other supplementary explanations of steps 201-206, please refer to the supplementary explanations of steps 101-106 in Embodiment 1. This embodiment of the invention will not repeat them here.

[0104] 207. Collect statistical data on work conditions and corresponding behavioral profiles generated within a preset time period to construct a group work dataset. The group work dataset is used to reflect the parameter configuration preferences of different operators in different work scenarios. 208. Input the group operation dataset into the preset hybrid decision model to update the historical baseline data in the multi-objective optimization operation; 209. When the behavior profile data of the target device deviates from the general characteristics in the group operation dataset by more than a preset threshold, the target function configuration parameters are calibrated according to the group operation dataset, and the calibrated target function configuration parameters are used as the recommended output.

[0105] In this embodiment of the invention, optionally, operational condition data and corresponding behavioral profile data generated within a preset time period are statistically analyzed to construct a group operation dataset. The group operation dataset reflects the parameter configuration preferences of different operators in different operational scenarios. The preset time period can be the past month or one thousand working hours, and the dataset records the parameter settings commonly used by most excellent drivers (filtered through profile data) under specific operating conditions.

[0106] The group task dataset is input into a pre-defined hybrid decision model to update the historical baseline data in the multi-objective optimization operation. The historical baseline data serves as a reference for the multi-objective optimization operation, and updating the baseline data means that the model has learned the latest and proven best operating practices.

[0107] When the behavioral profile data of a target device deviates from the common characteristics in the group operation dataset by more than a preset threshold, the target function configuration parameters are calibrated based on the group operation dataset. Deviations exceeding the threshold typically refer to the operating patterns of novice drivers. In this case, the system does not solely rely on the driver's individual profile but forcibly references the common optimized configuration in the group data. The calibrated target function configuration parameters are then used as recommended outputs to avoid equipment damage or inefficiency due to improper individual operation.

[0108] In this embodiment of the invention, optionally, the group operation dataset updates historical baseline data through a "clustering-weight transfer" mechanism. First, K-means clustering (e.g., K=50) is performed on the triples (operational condition data, behavioral profile data, parameter configuration) in the dataset to obtain a general parameter distribution for 50 typical operation scenarios. Second, the average performance of efficiency, energy consumption, and loss in each scenario is calculated as the historical baseline value for that scenario. Finally, the baseline value is injected into the multi-objective optimization engine of the hybrid decision model, replacing the original randomly initialized baseline weights, so that the model prioritizes learning the optimized configuration validated by the group.

[0109] In this embodiment of the invention, optionally, density peak clustering can be used for clustering of group task datasets to automatically determine the optimal number of clusters (without needing to preset a K value), thereby more accurately capturing the distribution characteristics of different task scenarios. Historical benchmark data updates support personalized adaptation: if the target device's behavioral profile data belongs to a group such as "Top 10% of proficiency," then only the benchmark values ​​for efficiency and energy consumption are updated; if it belongs to a "novice" group, then the benchmark values ​​for efficiency, energy consumption, and wear are updated simultaneously to prevent novices from imitating aggressive operations with high wear.

[0110] Further optional clustering methods for the group task dataset can employ various algorithms, such as partitioning clustering, density clustering, hierarchical clustering, or grid-based clustering. Historical benchmark data update strategies can include: direct replacement (overwriting the old benchmark with the new one), weighted averaging (proportionally merging the old and new benchmarks), or selective updates (updating only benchmarks with significant deviations). Personalized adaptation logic can also be diversified: such as adaptation based on profile tags (proficiency, aggressiveness), adaptation based on work condition type (heavy load, light load), or adaptation based on equipment status (new and old equipment).

[0111] Alternatively, the construction of the population dataset can incorporate data from multiple sources, such as equipment data, driver profile data, and task scheduling data. Baseline data can be injected in various ways, such as initializing model weights, using it as a regularization constraint in optimization, or serving as a reference baseline for the reward function.

[0112] Further optional steps for implementing dynamic evolution include: periodically deleting old data that exceeds a preset time threshold, adding newly collected job data, and maintaining the timeliness of the population dataset.

[0113] The implementation steps for federated learning are as follows: each device locally calculates the statistical features of the group dataset, uploads only the features to the cloud for aggregation, and then the cloud updates the global baseline data and distributes it to each device.

[0114] As can be seen, implementing this optional embodiment can construct a group operation dataset by statistically analyzing operational condition data and behavioral profile data within a preset time period. This enhances the aggregation capability of collective wisdom, thereby facilitating the extraction of excellent configuration experiences from different operators in different scenarios. This, in turn, provides a universal reference for individual decision-making, enabling the sharing and inheritance of group experience. The group operation dataset can update historical benchmark data in multi-objective optimization calculations, improving the timeliness and representativeness of the benchmark data. This helps reflect the mainstream configuration preferences in the current operational environment, guiding the model towards better group practices and achieving advanced parameter recommendations. Furthermore, by detecting deviations in individual behavioral profiles from the group's general characteristics, parameters can be calibrated, improving the fault tolerance for novice or abnormal operations. This helps avoid parameter configuration errors caused by insufficient individual experience, thereby reducing the risk of equipment damage due to improper operation and achieving both safety and economy in the operation process.

[0115] In another alternative embodiment, the method described above further includes: In response to the target device being in standby mode during a work interval, acquire the standby time of the target device and the current remaining work tasks; Based on standby time and remaining task data, predict the expected operating condition trend of the target equipment in the next operation phase; If the expected operating condition change trend indicates that an operating condition type switch is about to occur, the equipment operation protection logic that matches the expected operating condition change trend will be invoked in advance, and the priority allocation relationship of the multi-objective optimization operation in the preset hybrid decision model will be pre-adjusted to generate pre-adapted target function configuration parameters.

[0116] In this embodiment of the invention, optionally, in response to the target equipment being in a standby state during a work interval, the standby time of the target equipment and the current remaining work task data are obtained. The standby state during a work interval refers to the period of inactivity when the equipment stops operating but is not shut down. The remaining work task data can be obtained from the construction plan management system or calculated based on the currently completed workload.

[0117] Based on standby time and remaining tasks, predict the expected operating condition trend of the target equipment in the next operation phase. For example, if the standby time is short and the remaining tasks are mostly rock excavation, the next phase is predicted to be a heavy-load operation; if the standby time is long and it is close to shift change, the next phase is predicted to switch to a light-load or mobile operation.

[0118] If the expected trend of operating conditions indicates an impending switch in operating condition type, the system proactively invokes equipment operation protection logic that matches the expected trend and pre-adjusts the priority allocation of multi-objective optimization operations in the preset hybrid decision model to generate pre-adapted target function configuration parameters. For example, if it predicts an upcoming shift from excavation to long-distance travel, the system proactively deactivates the heavy-load protection logic and adjusts the priority to "speed priority," allowing it to directly output parameters adapted to the travel mode at the moment of operating condition switch, reducing response latency.

[0119] In this embodiment of the invention, optionally, the aforementioned pre-adjustment can be implemented through a "working condition switching probability-priority mapping table". The mapping table stores priority adjustment coefficients (ranging from 0.8 to 1.2) corresponding to 100 working condition switching types (such as "excavation → walking" and "loading → leveling"). For example, if it is predicted that the switching from "excavation" to "walking" will be imminent, the efficiency weight will be adjusted from 1.0 to 0.8 and the energy consumption weight from 1.0 to 1.2 according to the mapping table, so that the model can adapt to the low load requirements of the walking working condition in advance. The adjustment range is dynamically scaled according to the standby time (e.g., the longer the standby time, the larger the adjustment range, with a maximum of no more than 20%).

[0120] In this embodiment of the invention, optionally, the prediction of expected operating condition changes can also be combined with external data: such as weather forecast data (predicting the probability of rainfall). If the probability of rainfall is >70%, the priority is adjusted to "anti-sinking priority" (reducing hydraulic flow and increasing track ground pressure); or traffic condition data (predicting road congestion). If the probability of congestion is >60%, the priority is adjusted to "energy saving priority" (reducing engine idle speed). The triggering conditions for pre-adjustment can be further refined to: pre-adjustment is only initiated when the standby time is >5 minutes and the remaining workload is >30% of the current workload, to avoid over-response.

[0121] Further optionally, the prediction of expected operating condition change trends can incorporate multiple data sources: such as sensor data from the equipment itself, task data from the construction plan, weather / road condition data from the external environment, or historical operation pattern data. The pre-adjustment trigger conditions can be diverse: such as triggering based on standby time, triggering based on remaining workload, triggering based on the probability of operating condition switching, or triggering based on the operator's request. Priority adjustment methods can include: linear adjustment (proportional increase or decrease in weight), non-linear adjustment (adjustment based on the urgency of switching), or discrete adjustment (switching to a preset priority template).

[0122] Further, the prediction of expected changes in operating conditions can be based on various models, such as Markov chains, time series prediction models, or deep learning prediction models. The pre-adjustment can be implemented in conjunction with equipment status, such as engine temperature and hydraulic oil temperature, to ensure that the timing of adjustments aligns with the physical characteristics of the equipment.

[0123] Alternatively, the steps for implementing the cancellation mechanism are as follows: monitor whether the working condition switch is canceled; if it is canceled, restore the original parameter configuration; if the operator manually rejects the adjustment, terminate the pre-adjustment process.

[0124] The steps for short-term verification are as follows: continuously monitor the effect of parameter execution after the operating condition is switched, and if the actual performance deviates from the expectations, a second adjustment is triggered.

[0125] As can be seen, implementing this optional embodiment can improve the predictive ability of work condition switching by acquiring standby time and remaining task data in response to standby status. This facilitates the early detection of changes in work rhythm, allowing for a time window for parameter pre-adjustment and enabling forward-looking decision-making. By predicting expected work condition changes, the predictability of work condition switching is improved, which helps identify upcoming work condition type transitions and allows for advance preparation of parameter configurations adapted to the new work condition, achieving a smooth work transition. Furthermore, by pre-calling matching equipment operation protection logic and pre-adjusting priority allocation relationships, the response speed during work condition switching is improved, reducing parameter adaptation delays during switching and thus mitigating load impacts on equipment during transitions between old and new work conditions, ensuring continuity and stability of the work process.

[0126] Example 3 Please see Figure 3 , Figure 3 This is a schematic diagram of a device for recommending functional configuration parameters of construction machinery, as disclosed in an embodiment of the present invention. This device can be applied to construction machinery, such as excavators, and also to intelligent devices associated with construction machinery. These intelligent devices include, but are not limited to, one or more of the following: switching devices, cloud devices, edge computing devices, relay devices, base station devices, urban management devices, and intelligent connected devices. The present invention does not impose any limitations on these devices. Figure 3 As shown, the device for recommending the functional configuration parameters of this construction machinery may include: The data acquisition module 301 is used to acquire the operating data of the target equipment, the operating environment data, and the operating data of the target equipment's operating entity. The identification module 302 is used to identify the operating condition data of the target device based on the running data, the operating environment data and the preset time series deep learning model. The operating condition data is used to indicate at least the operating mode and the object of operation of the target device in the current environment. Module 303 is used to construct and dynamically update behavioral profile data that characterizes the behavioral features of the operating subject based on the operation data. The calculation module 304 is used to input the work condition data and behavior profile data into the preset hybrid decision model and output the target function configuration parameters. The preset hybrid decision model is used to perform multi-objective optimization calculations including work efficiency, energy consumption level and equipment loss, and has built-in equipment operation protection logic configured according to the work condition data. Execution module 305 is used to send the target function configuration parameters to the target device for execution; Update module 306 is used to update the preset hybrid decision model based on the feedback data after execution.

[0127] As can be seen, implementing the embodiments of the present invention can improve the comprehensiveness of data dimensions by collecting the operating data of the target equipment, the operating environment data, and the operating data of the operator, thereby facilitating the construction of a more complete equipment status view, which in turn improves the accuracy of subsequent decisions and enables intelligent adaptation of the equipment operation process. It can identify operating condition data through a preset temporal deep learning model, improving the dynamic adaptability of operating condition identification, thus helping to capture the continuous changing characteristics of equipment operation and avoiding the lag of static identification, achieving real-time and accurate operating condition judgment. It can perform multi-objective optimization calculations including operating efficiency, energy consumption level, and equipment wear through a preset hybrid decision model, improving the rationality of multi-objective trade-offs, thereby balancing the efficiency requirements and resource consumption of equipment operation, and thus reducing the overall cost of the equipment's entire life cycle, achieving sustainable operation. It can improve the accuracy of recommending functional configuration parameters for construction machinery and enhance the safety of equipment operation by incorporating equipment operation protection logic configured based on operating condition data, thereby helping to avoid the risk of damage to equipment due to exceeding parameters, and thus extending the equipment's service life, achieving a balance between operational safety and performance. It can update the preset hybrid decision model based on the feedback data after execution, thereby improving the model's dynamic evolution capability. This is conducive to adapting to long-term changes in equipment operating environment and operating habits, and thus improving the long-term effectiveness of parameter recommendations, realizing the model's self-optimization and continuous iteration.

[0128] In this embodiment of the invention, as an optional implementation, the specific method by which the identification module 302 identifies the operating condition data of the target device based on operating data, operating environment data, and a preset time-series deep learning model includes: Spatiotemporal alignment of operational data and work environment data yields a multidimensional feature sequence. The multidimensional feature sequence is input into the preset temporal deep learning model, and the state change features of the target device in a continuous time period are extracted through the self-attention mechanism to obtain the initial working condition data. Obtain a predefined work condition knowledge graph, which is a structured data set containing the logical relationships between work methods, work object attributes, work environment constraints, and target equipment work constraints; The initial operating condition data is mapped to the operating condition knowledge graph for matching, and the operating condition data that matches the logical relationship in the operating condition knowledge graph is filtered to determine the operating condition data of the target equipment.

[0129] As can be seen, implementing this optional embodiment can improve the temporal consistency of multi-source data by aligning operational data and work environment data in time and space, thereby helping to eliminate time deviations in data collection and facilitating the construction of coherent multi-dimensional feature sequences, thus ensuring the reliability of the data foundation. It can extract state change features within continuous time periods through the self-attention mechanism of a pre-defined temporal deep learning model, improving the depth and correlation of feature extraction, thereby helping to capture the evolutionary patterns of equipment actions and enhancing the representational ability of initial working condition data, enabling accurate depiction of complex work scenarios. It can match and filter working condition data through a predefined working condition knowledge graph, improving the logical rigor of working condition identification, thus helping to eliminate misidentification results that do not conform to physical common sense, ensuring the consistency between operational working condition data and real working scenarios, and ensuring the rationality of working condition judgments. It can improve the correlation between working condition data and work methods and work object attributes through the structured logical relationships of the working condition knowledge graph, thereby helping to clarify the parameter configuration boundaries under different work scenarios, providing a reliable basis for subsequent decision-making, and achieving scenario adaptability of parameter recommendations.

[0130] In this embodiment of the invention, as another optional implementation, the specific method by which the construction module 303 constructs and dynamically updates behavioral profile data to characterize the behavioral features of the operating subject based on the operation data includes: The first quantitative feature value is extracted from the operation data, which includes operation category, operation frequency, operation range and operation duration. The first quantitative feature value is used to indicate the intensity and duration of the operation subject under different operation actions. Behavioral profile data is determined based on the first quantitative feature value. The behavioral profile data includes at least one feature label among operational aggressiveness, efficiency preference, and proficiency. Based on the second quantitative feature value corresponding to the historical operation data, the feature labels in the behavior profile data are iteratively corrected, and the corrected behavior profile data is used as the current behavior profile data. The second quantitative feature value corresponds to the first quantitative feature value.

[0131] As can be seen, implementing this optional embodiment can improve the objectivity of behavioral feature quantification by extracting first quantitative feature values ​​of operation category, operation frequency, operation magnitude, and operation duration from operation data. This helps avoid the bias of subjective evaluation and, consequently, accurately reflects the intensity and duration of the operator's actions, thus establishing a data-driven foundation for behavioral profiling. It can determine behavioral profile data containing operational aggression, efficiency preferences, and proficiency based on the first quantitative feature values, improving the structured expression of operational habits. This helps transform implicit operational experience into explicit labels, providing personalized input features for decision-making models and enabling customized parameter recommendations. It can iteratively correct behavioral profile labels using second quantitative feature values ​​corresponding to historical operation data, improving the dynamic adaptability of behavioral profiles. This helps eliminate the impact of single abnormal operations, reflecting the long-term stable characteristics of the operator and ensuring the authenticity and reliability of profile data. A smooth update mechanism can fuse current and historical feature values, improving the stability of profile data. This helps avoid drastic changes in the profile due to short-term operational fluctuations, ensuring the consistency of parameter recommendations and achieving stable user experience.

[0132] In another optional implementation of this invention, the above-mentioned calculation module 304 inputs the work condition data and behavior profile data into the preset hybrid decision model, and outputs the target function configuration parameters in the following specific ways: Feature fusion is performed on operational condition data and behavioral profile data to obtain fused decision features; Based on the fusion decision characteristics, the operation strategy corresponding to the current operation scenario is determined by the operation condition data. The operation strategy is used to indicate the priority allocation relationship between operation efficiency, energy consumption level and equipment wear. Based on the operational strategy, multi-objective optimization is performed on at least one candidate parameter combination corresponding to the fusion decision features to obtain the comprehensive evaluation value of each candidate parameter combination; Based on the built-in equipment operation protection logic of the preset hybrid decision model, candidate parameter combinations that do not conform to the equipment operation protection logic are filtered out. The equipment operation protection logic is used to define the allowable range of parameters for the target equipment in the current operating environment. The parameter with the highest overall evaluation value among the remaining candidate parameter combinations after filtering is selected as the target function configuration parameter.

[0133] As can be seen, implementing this optional embodiment can improve the comprehensiveness of decision input by fusing features from operational condition data and behavioral profile data. This facilitates the integration of scenario features and user features, thereby enabling the construction of a joint "scenario-human" decision view and achieving two-dimensional adaptation of parameter recommendations. It can determine the operational strategy under the current operational scenario by fusing decision features, improving the dynamic adjustment capability of multi-objective priorities. This allows for flexible switching of the emphasis on efficiency, energy consumption, and loss according to operational needs, thus improving the contextual adaptability of parameter configuration and achieving optimal resource allocation. It can improve the comprehensiveness of the overall evaluation value by performing multi-objective optimization calculations on candidate parameter combinations based on operational strategies. This facilitates the evaluation of the overall performance of parameter combinations from multiple dimensions, enabling the selection of balanced solutions that take into account efficiency, energy consumption, and loss, and achieving collaborative optimization of multiple objectives. It can filter out candidate parameter combinations that do not meet the requirements through equipment operation protection logic, improving the safety of parameter output. This helps to eliminate invalid solutions that exceed physical limits or scenario limitations, ensuring that recommended parameters are within the safe operating range of the equipment and achieving controllable risks in the operational process.

[0134] In an optional embodiment, the specific method by which the update module 306 updates the preset hybrid decision model based on the feedback data after execution includes: Acquire feedback data generated during the execution of target function configuration parameters by the target device. The feedback data includes the operator's adjustment record of the target function configuration parameters and the actual operating deviation of the target device during the execution process. Based on the feedback data, determine the correction parameters of the preset hybrid decision model. The correction parameters are used to adjust the priority allocation relationship in the multi-objective optimization operation and / or the allowable range of parameters in the equipment operation protection logic. The hybrid decision model is updated based on the corrected parameters.

[0135] As can be seen, implementing this optional embodiment can improve the comprehensiveness of feedback information by acquiring feedback data including operational adjustment records and actual equipment operating deviations. This facilitates the capture of deviations between the actual effects of parameter execution and the expected goals, thereby improving the reliability of model correction and achieving precise closed-loop control. The feedback data can be used to determine correction parameters, improving the targeted direction of model adjustments. This allows for optimization of weak links in priority allocation relationships or equipment operation protection logic, thereby improving the model's adaptability to actual operating scenarios and enabling continuous optimization of parameter recommendations. Furthermore, updating the hybrid decision model based on correction parameters enhances the model's self-evolution capability, facilitating the transformation of practical operating experience into the model's intrinsic knowledge and reducing long-term deviations between the model and actual needs.

[0136] In another optional embodiment, the above-mentioned construction module 303 is also used to collect the work condition data and corresponding behavior profile data generated within a preset time period to construct a group work dataset. The group work dataset is used to reflect the parameter configuration preferences of different operators in different work scenarios. The update module 306 is also used to input the group operation dataset into the preset hybrid decision model in order to update the historical benchmark data in the multi-objective optimization operation; Optional, such as Figure 4 As shown, the device also includes: The calibration module 307 is used to calibrate the target function configuration parameters according to the group operation dataset when the behavior profile data of the target device deviates from the general characteristics in the group operation dataset by more than a preset threshold, and to use the calibrated target function configuration parameters as the recommended output.

[0137] As can be seen, implementing this optional embodiment can construct a group operation dataset by statistically analyzing operational condition data and behavioral profile data within a preset time period. This enhances the aggregation capability of collective wisdom, thereby facilitating the extraction of excellent configuration experiences from different operators in different scenarios. This, in turn, provides a universal reference for individual decision-making, enabling the sharing and inheritance of group experience. The group operation dataset can update historical benchmark data in multi-objective optimization calculations, improving the timeliness and representativeness of the benchmark data. This helps reflect the mainstream configuration preferences in the current operational environment, guiding the model towards better group practices and achieving advanced parameter recommendations. Furthermore, by detecting deviations in individual behavioral profiles from the group's general characteristics, parameters can be calibrated, improving the fault tolerance for novice or abnormal operations. This helps avoid parameter configuration errors caused by insufficient individual experience, thereby reducing the risk of equipment damage due to improper operation and achieving both safety and economy in the operation process.

[0138] In yet another alternative embodiment, optionally, such as Figure 4 As shown, the aforementioned device further includes: The acquisition module 308 is used to acquire the standby time and the current remaining task data of the target device in response to the target device being in a standby state during a work interval. The prediction module 309 is used to predict the expected operating condition change trend of the target equipment in the next operation stage based on the standby time and remaining task data. The generation module 310 is used to pre-call the equipment operation protection logic that matches the expected operating condition change trend if the expected operating condition change trend indicates that an operating condition type switch is about to occur, and to pre-adjust the priority allocation relationship of the multi-objective optimization operation in the preset hybrid decision model in order to generate pre-adapted target function configuration parameters.

[0139] As can be seen, implementing this optional embodiment can improve the predictive ability of work condition switching by acquiring standby time and remaining task data in response to standby status. This facilitates the early detection of changes in work rhythm, allowing for a time window for parameter pre-adjustment and enabling forward-looking decision-making. By predicting expected work condition changes, the predictability of work condition switching is improved, which helps identify upcoming work condition type transitions and allows for advance preparation of parameter configurations adapted to the new work condition, achieving a smooth work transition. Furthermore, by pre-calling matching equipment operation protection logic and pre-adjusting priority allocation relationships, the response speed during work condition switching is improved, reducing parameter adaptation delays during switching and thus mitigating load impacts on equipment during transitions between old and new work conditions, ensuring continuity and stability of the work process.

[0140] Example 4 Please see Figure 5 , Figure 5 This is a schematic diagram of the structure of another device for recommending functional configuration parameters of construction machinery disclosed in this embodiment of the invention. This device for recommending functional configuration parameters of construction machinery can be applied to construction machinery, such as excavators, and can also be applied to intelligent devices associated with construction machinery. These intelligent devices include, but are not limited to, one or more of the following: switching devices, cloud devices, edge computing devices, relay devices, base station devices, urban management devices, and intelligent connected devices. This embodiment of the invention does not impose any limitations. Figure 5 As shown, the device for recommending the functional configuration parameters of this construction machinery may include: Memory 401 that stores executable program code.

[0141] Processor 402 coupled to memory 401.

[0142] The processor 402 calls the executable program code stored in the memory 401 to execute the steps in the recommended method for the functional configuration parameters of engineering machinery described in Embodiment 1 or Embodiment 2 of the present invention.

[0143] Example 5 This invention discloses a computer storage medium storing computer instructions. When these computer instructions are invoked, they are used to execute the steps in the recommended method for configuring the functional parameters of engineering machinery as described in Embodiment 1 or Embodiment 2 of this invention.

[0144] Example 6 This invention discloses a computer program product, which includes a non-transitory computer storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps in the recommended method for configuring the functional parameters of engineering machinery as described in Embodiment 1 or Embodiment 2.

[0145] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0146] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0147] Finally, it should be noted that the method and apparatus for recommending functional configuration parameters of engineering machinery disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for recommending functional configuration parameters for engineering machinery, characterized in that, The method includes: Collect operational data of the target equipment, operating environment data, and operational data of the main operating entity of the target equipment; Based on the operational data, the working environment data, and the preset time-series deep learning model, the working condition data of the target device is identified, and the working condition data is used to at least indicate the working mode and working object of the target device in the current environment; Based on the operation data, construct and dynamically update behavioral profile data to characterize the behavioral features of the operation subject; The operation condition data and the behavior profile data are input into a preset hybrid decision model, and the target function configuration parameters are output. The preset hybrid decision model is used to perform multi-objective optimization calculations including operation efficiency, energy consumption level and equipment loss, and has built-in equipment operation protection logic configured according to the operation condition data. The target function configuration parameters are sent to the target device for execution. The preset hybrid decision model is updated based on the feedback data after execution.

2. The method for recommending functional configuration parameters of engineering machinery according to claim 1, characterized in that, The step of identifying the operating condition data of the target device based on the operating data, the working environment data, and a preset time-series deep learning model includes: The runtime data and the work environment data are spatiotemporally aligned to obtain a multidimensional feature sequence; The multidimensional feature sequence is input into a preset temporal deep learning model, and the state change features of the target device in a continuous time period are extracted through a self-attention mechanism to obtain initial working condition data. Obtain a predefined work condition knowledge graph, which is a structured data set containing the logical relationships between work methods, attributes of work objects, work environment restrictions, and equipment work restrictions of the target equipment; The initial operating condition data is mapped to the operating condition knowledge graph for matching, and the operating condition data that matches the logical relationship in the operating condition knowledge graph is filtered to determine the operating condition data of the target equipment.

3. The method for recommending functional configuration parameters of engineering machinery according to claim 1, characterized in that, The step of constructing and dynamically updating behavioral profile data to characterize the behavioral features of the operating subject based on the operational data includes: The operation data is used to extract a first quantitative feature value, which includes operation category, operation frequency, operation range and operation duration. The first quantitative feature value is used to indicate the intensity and duration of the operation subject under different work actions. Behavioral profile data is determined based on the first quantified feature value, and the behavioral profile data includes at least one feature label among operational aggressiveness, efficiency preference and proficiency. Based on the second quantized feature value corresponding to the historical operation data, the feature label in the behavior profile data is iteratively corrected, and the corrected behavior profile data is used as the current behavior profile data, wherein the second quantized feature value corresponds to the first quantized feature value.

4. The method for recommending functional configuration parameters of engineering machinery according to claim 1, characterized in that, The step of inputting the work condition data and the behavior profile data into a preset hybrid decision model and outputting target function configuration parameters includes: The operational condition data and the behavioral profile data are fused to obtain fused decision features; Based on the fusion decision features, the operation strategy corresponding to the operation condition data under the current operation scenario is determined. The operation strategy is used to indicate the priority allocation relationship between operation efficiency, energy consumption level and equipment wear. Based on the operational strategy, multi-objective optimization is performed on at least one candidate parameter combination corresponding to the fusion decision feature to obtain a comprehensive evaluation value for each candidate parameter combination. Based on the device operation protection logic built into the preset hybrid decision model, candidate parameter combinations that do not conform to the device operation protection logic are filtered out. The device operation protection logic is used to define the allowable range of parameters for the target device in the current operating environment. The parameter with the highest comprehensive evaluation value among the remaining candidate parameter combinations after filtering is determined as the target function configuration parameter.

5. The method for recommending functional configuration parameters of engineering machinery according to claim 4, characterized in that, The step of updating the preset hybrid decision model based on feedback data after execution includes: The feedback data generated during the execution of the target function configuration parameters by the target device is obtained. The feedback data includes the adjustment record of the target function configuration parameters by the operating subject and the actual operating deviation of the target device during the execution process. Based on the feedback data, the correction parameters of the preset hybrid decision model are determined. The correction parameters are used to adjust the priority allocation relationship in the multi-objective optimization operation and / or the allowable range of parameters in the equipment operation protection logic. The hybrid decision model is updated based on the corrected parameters.

6. The method for recommending functional configuration parameters of engineering machinery according to any one of claims 1-5, characterized in that, The method further includes: The work status data and corresponding behavior profile data generated within a preset time period are statistically analyzed to construct a group work dataset, which is used to reflect the parameter configuration preferences of different operating subjects in different work scenarios. The group operation dataset is input into the preset hybrid decision model to update the historical benchmark data in the multi-objective optimization operation; When the behavior profile data of the target device is detected to deviate from the common features in the group operation dataset by more than a preset threshold, the target function configuration parameters are calibrated according to the group operation dataset, and the calibrated target function configuration parameters are used as the recommended output.

7. The method for recommending functional configuration parameters of engineering machinery according to any one of claims 1-5, characterized in that, The method further includes: In response to the target device being in a standby state during a work interval, the standby time of the target device and the current remaining work tasks are obtained; Based on the standby time and the remaining task data, predict the expected operating condition change trend of the target equipment in the next operation stage; If the expected operating condition change trend indicates that an operating condition type switch is about to occur, the equipment operation protection logic that matches the expected operating condition change trend is invoked in advance, and the priority allocation relationship of the multi-objective optimization operation in the preset hybrid decision model is pre-adjusted to generate pre-adapted target function configuration parameters.

8. A device for recommending functional configuration parameters of engineering machinery, characterized in that, The device includes: The data acquisition module is used to collect the operating data of the target device, the operating environment data, and the operating data of the main body operating the target device; The identification module is used to identify the operating condition data of the target device based on the running data, the operating environment data and the preset time-series deep learning model. The operating condition data is used to indicate at least the operating mode and operating object of the target device in the current environment. The construction module is used to construct and dynamically update behavioral profile data that characterizes the behavioral features of the operating subject based on the operation data. The calculation module is used to input the work condition data and the behavior profile data into the preset hybrid decision model and output the target function configuration parameters. The preset hybrid decision model is used to perform multi-objective optimization calculations including work efficiency, energy consumption level and equipment loss, and has built-in equipment operation protection logic configured according to the work condition data. The execution module is used to send the target function configuration parameters to the target device for execution; The update module is used to update the preset hybrid decision model based on the feedback data after execution.

9. A device for recommending functional configuration parameters of engineering machinery, characterized in that, The device includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the recommended method for the functional configuration parameters of the engineering machinery as described in any one of claims 1-7.

10. A computer storage medium, characterized in that, The computer storage medium stores computer instructions, which, when invoked, are used to execute the recommended method for the functional configuration parameters of the engineering machinery as described in any one of claims 1-7.

Images