PI parameter scheduling method for improving engine torque control adaptability

Abstract

The invention discloses a PI parameter scheduling method for improving engine torque control adaptability, and relates to the technical field of engine electric control. The method is technically characterized in that a basic proportional integral parameter mapping table is established by taking the target rotating speed and the actual load of the engine as core dimensions, all running scenes of the engine from idle speed to the highest rated rotating speed and from no load to full load are comprehensively covered, and the target rotating speed and the actual load of the engine are calculated by combining sub-scene calibration and environmental factor compensation correction of a cold engine stage and a hot engine stage. It is ensured that parameters can be accurately matched with dynamic differences such as frictional resistance of an engine, response characteristics of a supercharger and fuel oil atomization effects under different working conditions; and secondly, engine body working condition adaptation and real-time control state adaptation are separated through two-dimensional decoupling, smooth transition of parameters in the working condition switching process is achieved through a multi-stage collaborative algorithm of rough matching, fine interpolation, boundary correction and transition smoothing, and control unbalance caused by parameter sudden change is avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of engine electronic control, in particular to a PI parameter scheduling method for improving the adaptability of engine torque control. BACKGROUND

[0002] As the core function of engine electronic control system, engine torque control directly determines the vehicle power output quality, operation equipment running stability and energy utilization efficiency, and is widely adapted to multiple application scenarios such as vehicle road driving, remote throttle precise control, power take-off (PTO) operation, and engineering machinery load driving. In the existing technical system, torque control is realized through torque control (TSC) message closed-loop regulation, and the proportional-integral (PI) controller becomes the mainstream core component of the closed-loop control due to its simple structure and low engineering implementation cost.

[0003] However, the running conditions of the engine are complex and changeable, and the intrinsic dynamic characteristics are significantly different under different speed and load conditions. For example, the engine is responsive under low load and high speed conditions, and the engine is slow to respond under high load and low speed conditions. At the same time, key parameters such as friction resistance, supercharger response delay, and fuel atomization characteristics also change with the working condition. The control scheme with fixed PI parameters cannot adapt to the above differences in dynamic characteristics, resulting in many problems in actual application.

[0004] First, the speed overshoots and drops. In the sensitive condition, the torque response overshoot is easily caused by excessive parameter adaptation, resulting in speed overshoot. In the slow condition, the torque response is slow due to insufficient parameter adaptation, causing the speed to drop.

[0005] Second, the running smoothness is poor. Fixed parameters are difficult to balance fast response and control stability, and oscillation is easily produced when the working condition is switched or the load is suddenly changed, affecting the running experience of the vehicle or the operation equipment.

[0006] Third, the adaptability is insufficient. The parameters are usually optimized only in a certain common working condition, and the control accuracy decreases in other working conditions, which cannot meet the performance requirements in all scenarios. Therefore, we propose a PI parameter scheduling method for improving the adaptability of engine torque control. SUMMARY

[0007] (I) Technical problems solved

[0008] In view of the shortcomings of the prior art, the present application provides a PI parameter scheduling method for improving the adaptability of engine torque control, which solves the technical problems of speed overshoot, slow response, poor smoothness and insufficient adaptability of the existing fixed PI parameter control scheme.

[0009] (II) Technical solutions

[0010] To achieve the above objectives, the present invention provides the following technical solution:

[0011] A PI parameter scheduling method for improving the adaptability of engine torque control, comprising the following steps:

[0012] A two-dimensional, hierarchical, decoupled adaptive gain scheduling system is constructed, consisting of an operating condition dimension and a deviation dimension. This system adapts to the dynamic characteristics of the engine body through the operating condition dimension and adapts to the real-time control state through the deviation dimension.

[0013] Establish a basic proportional-integral parameter mapping table, which takes the engine target speed and actual load as the core dimensions and covers all engine operating conditions.

[0014] The engine collects multi-dimensional operating signals, including target speed, actual speed, actual load, and dynamic characteristic signals of operating conditions related to torque control. After preprocessing the collected operating signals, the basic proportional-integral parameters corresponding to the current operating condition are obtained through the operating condition matching algorithm.

[0015] Calculate the engine speed control deviation and the rate of change of deviation, and construct a dynamic collaborative correction mechanism based on the two to correct the basic proportional-integral parameters; couple the basic proportional-integral parameters with the corresponding correction coefficients to synthesize the optimal proportional-integral parameters and perform safety verification.

[0016] The optimal proportional-integral (PI) parameters that have passed verification are input into the engine torque control system to generate torque control commands and send them to the actuators; the torque control effect is continuously monitored, and the basic PI parameter mapping table is iteratively optimized throughout its entire life cycle.

[0017] An integrated fault-tolerance mechanism is implemented to execute preset emergency handling strategies when a fault is detected, ensuring basic engine operation.

[0018] Preferably, the specific steps for establishing the basic proportional-integral parameter mapping table are as follows:

[0019] The mapping table structure is designed, including independent proportional gain basic mapping table and integral gain basic mapping table, both of which adopt a two-dimensional adaptation structure of target speed and actual load.

[0020] Set the target speed range to include the entire operating range of the engine from idle to maximum rated speed;

[0021] Choose the quantitative representation method of the actual load. You can choose any one of fuel injection quantity, intake air quantity or percentage load, and switch between different representation methods through calibration.

[0022] The systemic step response test under all operating conditions and in all environments was carried out on the engine bench, in two stages: cold engine and hot engine.

[0023] Multiple rounds of iterative optimization and calibration were performed on operating points with different speed and load combinations to determine the optimal basic parameters corresponding to each operating point; the parameters were compensated and corrected in combination with environmental factors, and the mapping table parameters were matched with the dynamic characteristics of the engine under various steady-state and transitional operating conditions.

[0024] Preferably, the specific steps for collecting multi-dimensional operating signals of the engine are as follows:

[0025] The types of signals collected include target speed, actual speed, actual load, and dynamic characteristic signals of operating conditions. Among them, the dynamic characteristic signals of operating conditions cover temperature, pressure, load change rate, and throttle opening.

[0026] High-precision sensors adapted to various signal detection requirements are selected to form a sensor group; and various signals are synchronously collected through the sensor group at a frequency consistent with the engine control cycle.

[0027] The collected data are processed sequentially, including outlier handling, noise reduction, synchronization alignment, and normalization preprocessing.

[0028] Preferably, the specific steps for obtaining the basic proportional-integral parameters through operating condition matching are as follows:

[0029] Based on the real-time target speed and actual load, the engine is positioned to a preset operating range, which is divided according to the engine's dynamic characteristic change point.

[0030] The parameters of the mapping table within the working condition interval are calculated using an interpolation algorithm to obtain preliminary basic proportional-integral parameters;

[0031] If the current operating point is close to the boundary of the mapping table, the boundary constraint correction strategy is used to adjust the initial basic parameters; if the engine is in the process of switching operating conditions, the transition smoothing processing strategy is used to optimize the initial basic parameters to control the torque output to be stable and without oscillation.

[0032] Preferably, the specific steps for calculating the speed control deviation and the rate of change of deviation and constructing a dynamic correction mechanism are as follows:

[0033] Calculate the real-time difference between the engine target speed and the actual speed to obtain the speed control deviation;

[0034] The speed control deviation is processed using a preset signal preprocessing method to suppress instantaneous fluctuation interference; the quantified value of the speed deviation change trend within a continuous control cycle is calculated to obtain the speed deviation change rate;

[0035] The transient and steady-state conditions of the operating conditions are identified by the rate of change of speed deviation; the speed control deviation and the rate of change of deviation are set as dual input variables for dynamic correction, and a dynamic collaborative correction mechanism is constructed.

[0036] Preferably, the specific steps for correcting the basic proportional-integral parameters are as follows:

[0037] A dynamic correction system is constructed, which includes proportional gain correction coefficient and integral gain correction coefficient;

[0038] The logic for setting the value of the correction coefficient is set and dynamically adjusted based on the bivariate coupling relationship between the absolute value of the speed deviation and the rate of change of the deviation;

[0039] Based on the differences in combustion characteristics between spark-ignition and compression-ignition engines, the correction coefficients are calibrated differently.

[0040] If it is a transient process, the correction coefficient is adjusted to enhance the proportional effect and weaken the integral effect; if it is a steady-state process, the correction coefficient is adjusted to weaken the proportional effect and enhance the integral effect; if it is a transient process, the correction coefficient is dynamically adjusted according to the linear transient logic.

[0041] The basic proportional-integral parameters are corrected using the adjusted correction factors.

[0042] Preferably, the specific steps for synthesizing the optimal proportional-integral parameters and performing security checks are as follows:

[0043] The basic proportional-integral parameters are coupled with the corresponding correction coefficients to obtain the preliminary optimal proportional-integral parameters;

[0044] Perform a safety range check on the preliminary optimal proportional-integral parameters, eliminate parameters that exceed the preset safety boundary, and limit the rate of parameter change;

[0045] The above coupling and verification process is executed in a closed loop within each engine control cycle, and the control parameter response is synchronized with the changes in operating conditions in real time.

[0046] Preferably, the specific steps for integrated proportional-integral scheduling and feedforward compensation coordinated control are as follows:

[0047] By monitoring the engine operating status through the rate of load change, scenarios of sudden load changes can be identified.

[0048] The preset multi-dimensional feedforward compensation mapping table is invoked, and the appropriate torque compensation value is retrieved based on the current target speed, actual load, and load change range.

[0049] A dynamic weighting strategy is set up to increase the weight of the feedforward compensation value in the transient process and decrease the weight of the feedforward compensation value in the steady-state process.

[0050] The feedforward compensation value is superimposed with the torque correction command output by the proportional-integral controller according to the set weight to generate the final torque control command and send it to the actuator.

[0051] Preferably, the specific steps for iterative optimization of the basic proportional-integral parameter mapping table are as follows:

[0052] Set core evaluation indicators, including torque control deviation rate, response time, overshoot, and number of oscillations;

[0053] During actual engine operation, parameter application data and corresponding control effect data are continuously recorded according to operating point classification.

[0054] The optimization judgment is triggered according to a preset cycle. If the control effect of a certain working point is consistently better than the initial calibration value and has been verified as qualified by real vehicles in multiple environments, the mapping table parameters of that working point will be automatically updated.

[0055] Set up a parameter backtracking mechanism so that if the updated parameters cause a decrease in control performance, the system will automatically backtrack to the historical optimal value.

[0056] When the cumulative updated operating point reaches the preset ratio, batch calibration is initiated to adjust the correlation of parameters of adjacent operating points.

[0057] Preferably, the execution and adaptation steps of this method are as follows:

[0058] The engine electronic control unit serves as the execution carrier. This electronic control unit integrates a fault-tolerant function module. The fault-tolerant module has a preset emergency handling strategy for scenarios such as abnormal sensor data, parameter calculation failure, and communication failure. When a fault is detected, it switches to emergency proportional-integral parameters within 10ms to control the engine to maintain basic operation.

[0059] It is compatible with spark-ignition and compression-ignition engines of various displacements and cylinder numbers.

[0060] (III) Beneficial Effects

[0061] 1. A basic proportional-integral (PI) parameter mapping table, built around the engine target speed and actual load as core dimensions, comprehensively covers all engine operating scenarios from idle to maximum rated speed and from no load to full load. Combined with scenario-specific calibration and environmental factor compensation correction during cold and warm engine phases, it ensures that parameters can accurately match the dynamic differences in engine friction resistance, turbocharger response characteristics, and fuel atomization effect under different operating conditions. 2. A dual-dimensional decoupling separates engine operating condition adaptation from real-time control state adaptation. Then, through a multi-level collaborative algorithm of coarse matching, fine interpolation, boundary correction, and transition smoothing, it achieves a smooth transition of parameters during operating condition switching, avoiding control imbalance caused by parameter mutations. Finally, the PI parameters can be dynamically adjusted according to changes in engine operating conditions, perfectly adapting to various scenarios such as cold start, idle warm-up, high-speed cruising, and full-load acceleration, solving the technical problem of insufficient adaptability to all operating conditions in existing technologies.

[0062] 2. The dynamic collaborative correction mechanism accurately identifies the transient, steady-state, and transient states of the operating conditions and adjusts the correction coefficients of the proportional gain and integral gain differently. Specifically, during the transient process, the proportional action is strengthened to quickly eliminate control deviations, while the integral action is weakened to avoid overshoot. During the steady-state process, the proportional action is weakened to ensure operational stability, while the integral action is strengthened to completely eliminate steady-state error, thus achieving a balance between response speed and stability from a control logic perspective. The feedforward compensation module monitors load step changes in advance and predictively outputs torque compensation values, which are dynamically superimposed on the proportional-integral control output. This effectively compensates for the response lag problem of pure feedback control, enabling the engine to maintain stable speed even under dynamic scenarios such as sudden load changes, and completely improving the speed surge and drop phenomena commonly found in existing technologies.

[0063] 3. In the operating condition matching stage, the boundary constraint correction strategy avoids abnormal fluctuations in parameters due to their proximity to the mapping table boundaries. The transition smoothing process ensures the stability of torque output during operating condition switching by suppressing step changes in parameters, completely eliminating the oscillation and jitter problems that are prone to occur in fixed parameter schemes. In the parameter synthesis stage, multi-layered mechanisms such as safety range verification, rate of change limitation, and special operating condition rationality verification effectively prevent control shocks caused by abnormal parameters. Furthermore, the full life cycle iterative optimization based on quantitative evaluation of control effect continuously records the parameter application effect at each operating point, automatically updates the optimal parameters, and ensures the consistency of parameters at adjacent operating points. This enables the engine to maintain a highly accurate and smooth control state throughout the entire use process, avoiding the problem of control performance degradation after long-term use. Attached Figure Description

[0064] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0065] Figure 1 This is a flowchart of an embodiment of the present invention. Detailed Implementation

[0066] This application provides a PI parameter scheduling method to improve the adaptability of engine torque control, solving the technical problems of speed overshoot, slow response, poor smoothness, and insufficient adaptability in existing fixed PI parameter control schemes. The basic proportional-integral parameter mapping table, built with the engine target speed and actual load as the core dimensions, comprehensively covers all operating scenarios of the engine from idle speed to maximum rated speed and from no load to full load. Combined with scenario-specific calibration and environmental factor compensation correction during cold and hot engine stages, it ensures that the parameters can accurately match the dynamic differences in engine friction resistance, turbocharger response characteristics, fuel atomization effect, etc. under different operating conditions.

[0067] Example: The technical solution in this application aims to solve the technical problems of speed overshoot, slow response, poor smoothness, and insufficient adaptability in existing fixed PI parameter control schemes. The overall approach is as follows:

[0068] To address the problems existing in the prior art, this invention provides a PI parameter scheduling method for improving the adaptability of engine torque control. The scheduling method comprises the following steps:

[0069] To address the technical issues of poor adaptability and control performance imbalance of traditional fixed proportional-integral parameters under various operating conditions, including engine cold start, idling, partial load, full load, operating condition switching, and sudden load changes, a two-dimensional, hierarchically decoupled adaptive gain scheduling system combining operating condition and deviation dimensions is constructed to synergistically optimize the torque control response speed, stability, and accuracy under all engine operating conditions. Details are as follows:

[0070] Firstly, a dual-dimensional, layered, decoupled adaptive gain scheduling system is employed. This system comprises two parts: a condition-dimensional adaptation unit and a deviation-dimensional adaptation unit. The condition-dimensional adaptation unit adapts to the dynamic characteristics of the engine itself, using target speed and actual load as adaptation dimensions. It outputs basic parameters matching the current operating condition through a pre-calibrated basic proportional-integral parameter mapping table. Simultaneously, it integrates an environmental parameter compensation interface, enabling real-time reception of external environmental signals such as ambient temperature and atmospheric pressure to perform preliminary environmental corrections on the basic parameters, eliminating the influence of the external environment on parameter adaptability. Secondly, the deviation-dimensional adaptation unit dynamically adapts to the engine's real-time control state. Using speed control deviation and deviation change rate as input data, it outputs corresponding dynamic correction coefficients through a built-in condition state recognition algorithm. This unit possesses accurate identification capabilities for transient, steady-state, and transient operating states, and can dynamically adjust the value logic of the correction coefficients according to the actual control state of the engine, ensuring that the corrected parameters accurately match the real-time control requirements.

[0071] In the above, the basic parameters output by the working condition dimension adaptation unit serve as the input benchmark for the deviation dimension adaptation unit. The deviation dimension adaptation unit dynamically calibrates the basic parameters based on the correction coefficient output by the real-time control status. After safety verification, the calibrated parameters are output to the torque control system. At the same time, the calibrated optimal parameters are fed back to the mapping table iterative optimization module of the working condition dimension, realizing the closed-loop flow of dual-dimensional data.

[0072] The full-process cyclic control specifically includes: constructing a two-dimensional, layered, decoupled adaptive gain scheduling system, completing the functional definition, parameter interface, and cooperative logic solidification of the two adapting units; establishing a basic proportional-integral parameter mapping table, completing bench calibration and environmental compensation correction for all operating conditions; multi-dimensional operation signal acquisition, covering the comprehensive acquisition of core control signals and dynamic characteristic signals of operating conditions; signal preprocessing to improve the quality and reliability of acquired signals; precise operating condition matching, locating the current operating condition and outputting basic parameters through multi-level algorithms; dynamic cooperative correction, accurately calibrating basic parameters based on real-time control status; optimal parameter synthesis and safety verification to ensure the safety and real-time performance of output parameters; cooperative control command generation and execution to achieve precise torque control; full lifecycle iterative optimization of the mapping table to ensure long-term stable operation performance; and a fault-tolerant mechanism to ensure operational reliability throughout the process.

[0073] Multi-mechanism collaboration is the core of achieving control performance optimization. Among them, the operating condition matching mechanism ensures the adaptability of basic parameters to operating conditions, the dynamic correction mechanism ensures the tracking of parameters to real-time control status, the iterative optimization mechanism ensures the stability of parameters in long-term operation, and the fault tolerance mechanism ensures the operational safety under extreme scenarios.

[0074] In practical implementation, a two-dimensional, layered, decoupled adaptive gain scheduling system consisting of operating condition and deviation dimensions is first constructed: First, an operating condition dimension adaptation unit is built, defining the quantization range of the target speed (range: idle speed - rated speed) and actual load (range: 0 - maximum load), along with basic parameter query and environmental correction logic; then, a deviation dimension adaptation unit is built, defining the calculation logic for speed control deviation (target speed - actual speed) and deviation change rate (difference in deviation between adjacent cycles / control cycle), including transient (deviation change rate > 50 r / (min・cycle)) and steady-state (deviation change rate < 10 r / (min・cycle)) conditions. The identification thresholds for the cycle and transition state (10r / (min・cycle) ≤ deviation change rate ≤ 50r / (min・cycle)) are set; the collaborative logic of the two units is defined, that is, the basic parameters output by the operating condition dimension are used as the input benchmark for the deviation dimension, the correction coefficient output by the deviation dimension is used to calibrate the basic parameters, the calibrated parameters are output after safety verification, and the optimal calibrated parameters are fed back to the mapping table iteration module of the operating condition dimension. The collaborative logic is solidified into the algorithm module of the engine electronic control unit through C language code; then, a basic proportional integral parameter mapping table is established to complete the full operating condition point mapping. Calibration and correction provide a reliable source of basic parameters for operating condition matching. Then, core engine control signals and dynamic operating condition characteristic signals are collected, including coolant temperature, intake pressure, load change rate, fuel temperature, exhaust temperature, throttle opening, atmospheric pressure, and ambient temperature. A full-process preprocessing, including outlier removal, noise reduction, synchronization alignment, and normalization, is performed to provide highly reliable data for subsequent algorithms. Subsequently, a four-level collaborative algorithm—coarse matching, fine interpolation, boundary correction, and transition smoothing—completes accurate operating condition matching, outputting the basic proportional-integral parameters corresponding to the current operating condition. Finally, the original speed control deviation is calculated and eliminated through sliding average preprocessing. Excluding instantaneous disturbances, the deviation change rate is calculated and the current operating condition is identified. Based on bivariate coupling logic, the correction coefficient is dynamically adjusted to complete the dynamic correction of the basic proportional-integral parameters. Then, the basic parameters and correction coefficients are coupled to synthesize the preliminary optimal parameters. Multi-layer safety checks are then performed sequentially, including safety range verification, parameter change rate limitation, and special operating condition rationality verification, to output qualified final optimal proportional-integral parameters. The optimal proportional-integral parameters are then input into the engine torque control system, integrating a feedforward compensation module to identify load step scenarios and retrieve the corresponding compensation values. The final torque control command is generated through dynamic weight superposition and transmitted via CAN2.The 0 bus sends torque control messages to the engine actuators (cycle: 10ms). During actual engine operation, the basic proportional-integral parameter mapping table undergoes full lifecycle iterative optimization. An optimization judgment is triggered every 100 working cycles, updating the parameters at the operating points that meet the conditions. Batch calibration is initiated when the cumulative updated operating points reach 30% of the total number of points. Throughout the torque control process, a fault tolerance mechanism is integrated, pre-setting typical fault scenarios such as abnormal sensor data, parameter calculation failures, and communication failures. Corresponding emergency proportional-integral parameters are preset for each type of fault, and fault detection thresholds are set to ensure automatic switching to emergency parameters within 10ms after a fault is detected. Simultaneously, the instrument fault indicator light illuminates and the fault code is stored, achieving rapid fault detection, emergency response, and information recording.

[0075] 1. Basic proportional-integral parameter mapping table:

[0076] To address the issues of insufficient compatibility and low accuracy between traditional mapping table parameters and engine dynamic characteristics, this method aims to ensure a reliable and accurate source of basic proportional-integral parameters for the operating condition matching process, while also guaranteeing the real-time nature of parameter calls.

[0077] We employ an integrated construction logic that integrates a dual-table independent structure, full-condition coverage, phased bench calibration, and environmental compensation correction. Each core element supports the others, forming a closed loop. Specifically, the dual-table independent structure involves constructing two independent proportional gain and integral gain base mapping tables. Both use a two-dimensional adaptation structure based on target speed and actual load. The row dimension of the mapping table represents the target speed, divided with uniform steps, covering the entire range from idle to rated speed. The column dimension represents the actual load, divided with uniform steps, covering the entire range from no load to full load. This structure avoids mutual interference between proportional and integral gain parameters. Proportional gain primarily affects torque response speed, while integral gain primarily affects steady-state deviation. Independent optimization of both can improve response performance and steady-state accuracy respectively. Simultaneously, the mapping tables use a matrix storage structure stored in the Flash memory of the engine electronic control unit, supporting fast index lookups by target speed and actual load, ensuring real-time parameter retrieval.

[0078] Full operating condition coverage is the core of the mapping table, and it must meet three conditions: First, complete speed coverage, ensuring that the target speed range includes all possible operating speeds of the engine without any speed gaps; second, complete load coverage, ensuring that the actual load range includes the load range of the engine in all application scenarios such as vehicle drive, power take-off operation, and construction machinery operation; and third, reasonable operating condition point density, with step size settings balancing the mapping table's storage capacity and parameter accuracy, with speed step size not exceeding 50 r / min and load step size not exceeding 5% of the maximum load, ensuring smooth parameter changes between adjacent operating condition points and avoiding sudden parameter changes during operating condition switching.

[0079] Phased bench calibration is a key method to ensure the accuracy of the mapping table parameters. The calibration process is carried out separately in the cold engine stage and the warm engine stage. The cold engine stage corresponds to the start-up and warm-up operation state with engine coolant temperature of -40℃ to 60℃, and the warm engine stage corresponds to the normal operation state with engine coolant temperature of 80℃ to 100℃. Phased testing can ensure the adaptability of the mapping table parameters under different temperature conditions. The bench test adopts the step response test method. The specific process is as follows: fix the engine on the bench, apply a specified load through the bench dynamometer, control the engine to reach the target speed and run stably for 30 seconds; input the torque step command to the actuator through the engine electronic control unit; record the engine torque response curve and speed change curve in real time through the bench data acquisition system; adjust the proportional-integral parameters according to the evaluation index of the response curve, repeat the test and adjustment process until the index is met, and record the parameters at this time as the initial basic parameters for this operating point.

[0080] The environmental compensation and correction stage is used to eliminate the influence of environmental factors on parameter adaptability. These environmental factors mainly include ambient temperature and atmospheric pressure. By simulating different combinations of ambient temperature and atmospheric pressure on a test bench, the parameter deviations at the same operating point under different environmental conditions are tested. The variation law of deviation with ambient temperature and atmospheric pressure is analyzed, and an environmental compensation function is fitted to make targeted corrections to the basic parameters obtained from the test bench calibration. For example, when the ambient temperature is below -20℃, the proportional gain needs to be corrected and increased by 5%-8% to enhance the low-temperature start-up response; when the atmospheric pressure is below 90kPa, the integral gain needs to be corrected and increased by 3%-5% to compensate for steady-state deviations in high-altitude environments.

[0081] In practical implementation, the overall structure of the mapping table is first determined: an independent proportional gain base mapping table and an integral gain base mapping table are constructed. The range of values ​​for the row dimension (target speed) and column dimension (actual load) of the two mapping tables is clarified. Operating points are divided according to a speed step of 50 r / min and a load step of 5% of the maximum load. The matrix size of the mapping table is then determined. For example, from idle speed 750 r / min to rated speed 4500 r / min, there are 76 speed operating points; from load 0% to 100%, there are 21 load operating points. The mapping table size is 76×21, stored in matrix form. Next, the target speed coverage range is set: starting from the engine idle speed and ending at the rated speed, with uniform steps of 50 r / min. The system divides the operating speed points, assigning a unique index to each point, such as index 0 for 750 r / min, index 1 for 800 r / min, and so on. Then, it selects a quantification method for the actual load: percentage load (actual load / maximum load × 100%) is used as the quantification method, starting at 0% and ending at 100%, with load operating points divided in uniform 5% increments. Each load operating point is assigned a unique index, such as index 0 for 0% and index 1 for 5%, and so on. Simultaneously, an interface is reserved for switching between fuel injection quantity and intake air quantity load characterization methods, enabling one-click switching via the parameter configuration function of the calibration software. After switching, the column dimensions of the mapping table are automatically updated to the corresponding characterization method's operating point.

[0082] Subsequently, a systematic step response test was conducted on the bench under full operating conditions and in all environments: First, environmental pretreatment was performed, and the temperature and atmospheric pressure of extreme and standard environments were set in the bench environmental simulation chamber; then, the test was carried out in two stages: cold start and hot start. In the cold start stage, the test temperature was set in increments of 10°C from -40°C to 60°C, and in the hot start stage, the test temperature was set in increments of 5°C from 80°C to 100°C. All operating points were traversed point by point according to the pre-set speed and load step size. For each operating point, the step response test procedure was followed to complete the test and parameter adjustment, and the initial basic parameters after reaching the standard were recorded; the parameters of each operating point were tested under different environmental conditions, and the parameter deviation data were recorded.

[0083] Then, at least three rounds of iterative optimization and calibration were performed for each operating point: the first round was to initially adjust the parameters to ensure that there was no significant overshoot in the torque response; the second round was to optimize the parameters in a targeted manner to minimize the torque response time; the third round was to fine-tune the parameters to minimize the steady-state deviation; after each round of adjustment, the corresponding torque response curve, speed change curve and parameter values ​​were recorded to form a complete calibration log, and finally the optimal basic proportional-integral parameters for each operating point under standard conditions were determined.

[0084] Finally, an environmental compensation model is established: based on parameter deviation data under different ambient temperatures and atmospheric pressures, the least squares method is used to fit the environmental compensation function, and the parameter correction coefficients under different environmental conditions are determined; the basic proportional-integral parameters of each operating point are corrected according to the environmental compensation function, and the final proportional gain basic mapping table and integral gain basic mapping table are generated; the two mapping tables are written to the Flash memory of the engine electronic control unit through a calibration tool to complete the construction of the mapping tables.

[0085] 2. Acquisition and preprocessing of multi-dimensional operating signals:

[0086] The acquisition and preprocessing of multi-dimensional operating signals solves the problem of unreliable algorithm input data caused by sensor noise, electromagnetic interference, timing inconsistencies, and inconsistencies in units of measurement in the original acquired signals. Through multi-dimensional signal acquisition and full-process preprocessing, it provides highly reliable and high-quality data support for all subsequent modules such as working condition matching and dynamic correction.

[0087] The integrated processing logic, encompassing multi-dimensional signal selection, high-precision sensor configuration, synchronous acquisition mechanisms, and multi-stage preprocessing, ensures the comprehensiveness, accuracy, temporal consistency, and standardization of data through the synergistic effect of each core element. Multi-dimensional signal selection is crucial for ensuring the comprehensiveness of the acquired data. Acquired signals include two categories: core control signals and dynamic characteristic signals of operating conditions. Core control signals are fundamental and critical for ensuring engine torque control, directly participating in the calculation and adjustment of proportional-integral parameters. These include target speed, actual speed, and actual load: target speed is the command signal received by the engine electronic control unit from the host computer (such as the vehicle controller or remote throttle controller), representing the required engine speed level; actual speed is the real-time operating speed of the engine crankshaft, directly measured by a speed sensor, representing the actual operating state of the engine; actual load is the current actual load on the engine, measured by a dynamometer or estimated based on intake air volume and fuel injection quantity. The calculated load characteristics represent the engine's operating conditions. The dynamic characteristic signals of the operating conditions are auxiliary signals that comprehensively characterize the engine's operating conditions and environmental conditions. They are used to optimize the accuracy of operating condition matching and correct for the influence of the environment on parameters. These include coolant temperature, intake pressure, load change rate, fuel temperature, exhaust temperature, throttle opening, atmospheric pressure, and ambient temperature. Coolant temperature characterizes the thermal state of the engine body, affecting its combustion efficiency and dynamic characteristics. Intake pressure characterizes the working state of the engine's intake system, affecting fuel injection quantity and torque output. Load change rate characterizes the trend of load changes, used to identify load step scenarios. Fuel temperature affects fuel atomization, thus affecting the combustion process. Exhaust temperature characterizes the engine's combustion state, used to monitor for overheating risks. Throttle opening characterizes the control state of intake air volume, directly affecting engine power output. Atmospheric pressure and ambient temperature characterize external environmental conditions, used for environmental compensation and correction.

[0088] High-precision sensor configuration is fundamental to ensuring signal acquisition accuracy. Appropriate high-precision sensors are selected for different types of acquired signals: Hall effect speed sensors are installed at the signal disc at the front of the engine crankshaft, outputting speed pulse signals by detecting changes in the grooves of the signal disc; piezoresistive intake pressure sensors are installed on the engine intake manifold, directly measuring the pressure within the intake manifold; NTC coolant temperature sensors are installed at the engine block water jacket, in direct contact with the coolant; photoelectric throttle position sensors are integrated inside the throttle body, detecting the throttle opening position; electromagnetic fuel flow sensors are installed in the fuel line, measuring the instantaneous fuel flow rate; NTC fuel temperature sensors are installed at the fuel rail; NTC exhaust temperature sensors are installed at the exhaust manifold outlet; piezoresistive atmospheric pressure sensors are installed in a location within the engine compartment free from airflow interference; and NTC ambient temperature sensors are installed outside the engine compartment, measuring the external ambient temperature. All sensors must possess electromagnetic interference immunity and comply with the ISO 11452 electromagnetic compatibility standard.

[0089] The synchronous acquisition mechanism is crucial for ensuring the timing consistency of various signals. The signal acquisition frequency is kept completely consistent with the engine control cycle, ensuring that a complete set of signal data can be acquired in each control cycle for parameter calculation and control decisions. Based on the internal clock of the engine electronic control unit, a corresponding timestamp is added to the signal acquired by each sensor. The timestamp format is "control cycle number + acquisition time", ensuring that the timestamps of all signals within the same control cycle are consistent. For analog signals, they are converted into digital signals by the built-in AD converter of the engine electronic control unit. For digital signals, such as speed pulses and throttle opening digital signals, they are directly received and processed through the digital signal input interface. All acquired signals are temporarily stored in the RAM buffer of the engine electronic control unit to ensure that data is not lost.

[0090] A multi-stage preprocessing process is crucial for improving the quality of acquired data. The entire preprocessing process includes four stages: outlier removal, noise reduction, synchronization alignment, and normalization. Each stage progresses sequentially, gradually optimizing data quality. The outlier removal stage removes abnormal data from the original signal caused by sensor malfunctions, poor wiring connections, or strong external interference. It employs the 3σ criterion: calculating the average value μ and standard deviation σ of each signal over the most recent 10 control cycles. If the signal value in the current cycle exceeds the range [μ-3σ, μ+3σ], it is considered an outlier and replaced with the normal signal value from the previous cycle. The abnormal state is also marked for fault diagnosis. The noise reduction stage suppresses signal fluctuations caused by electromagnetic interference, sensor noise, and other factors. Corresponding noise reduction algorithms are used for different signal types: for high-frequency fluctuating signals such as speed and pressure, a Kalman filter algorithm is used, with a linear state equation and a nonlinear observation equation. The process noise variance Q=0.01 and the observation noise variance R=0. .05; For low-frequency, slowly changing signals such as temperature, a moving average filtering algorithm is used with a window size of 5 control cycles. The average value of the signal within the window is taken as the filtered value. The synchronization alignment stage is used to correct the timing deviation between different signals. Based on the timestamp of the actual speed signal, the timestamps of other signals are aligned with the speed signal. For signals with inconsistent timestamps, linear interpolation is used to supplement missing data or correct deviation data to ensure that the timing of all signals within the same control cycle is completely consistent. The normalization processing stage is used to unify the numerical range of signals with different dimensions to avoid algorithm calculation errors caused by excessive differences in signal amplitude. All signal values ​​are mapped to the normalization interval [0,1]. The normalization formula is: Normalized value = (current signal value - minimum signal value) / (maximum signal value - minimum signal value). The maximum and minimum values ​​of the signal are determined according to the measurement range of the sensor. For example, if the speed sensor has a measurement range of 0-6000 r / min, the maximum value = 6000 and the minimum value = 0.

[0091] In practical implementation, first, clarify the types of signals to be collected: determine the core control signals as target speed, actual speed, and actual load, and the dynamic characteristic signals of operating conditions as coolant temperature, intake pressure, load change rate, fuel temperature, exhaust temperature, throttle opening, atmospheric pressure, and ambient temperature. List the names, units, measurement ranges, and accuracy requirements of all signals. Next, configure and install the corresponding sensors according to the signal type: fabricate and install sensor brackets according to the engine's structural dimensions and installation location; install the speed sensor at the signal disc at the front end of the engine crankshaft, ensuring a gap of 0.5-1.0 mm between the sensor and the signal disc; install the intake pressure sensor on the engine intake manifold via a threaded interface. Secure the sensor to ensure a leak-free seal; install the coolant temperature sensor in the engine block water jacket, immersing it in the coolant, ensuring a firm fit; integrate the throttle position sensor inside the throttle body and connect it to the throttle shaft; install the fuel flow sensor and fuel temperature sensor on the fuel rail to ensure smooth fuel flow; install the exhaust temperature sensor at the exhaust manifold outlet, secured with a high-temperature resistant bracket; install the atmospheric pressure sensor and ambient temperature sensor in their designated locations; all sensor wiring harnesses are shielded, with the shield grounded to reduce electromagnetic interference; perform precise calibration of all sensors using professional equipment, following the sensor calibration procedures to ensure the measurement accuracy of all signals meets standards.

[0092] Next, set the signal acquisition parameters and synchronization mechanism: set the signal acquisition frequency to 10ms to match the engine control cycle; configure the AD converter parameters of the engine electronic control unit, setting the resolution to 12 bits; enable the timestamp function based on the internal clock of the engine electronic control unit, adding a timestamp to each acquired signal; configure the signal input interface, connecting analog signals to the AD converter channel and digital signals to the digital input interface; temporarily store all acquired signals in the RAM buffer of the engine electronic control unit, setting the write and read rules for the buffer to avoid data overwriting or loss.

[0093] Finally, the collected raw data undergoes a full-process preprocessing: First, outlier removal is performed by calculating the average and standard deviation of the last 10 cycles for each signal to determine if the current signal value is an outlier. If it is, it is replaced and marked. Next, noise reduction is performed by applying Kalman filtering to signals such as engine speed and intake pressure, and moving average filtering to temperature signals to obtain the noise-reduced signal. Then, synchronization alignment is performed by adjusting the timestamps of other signals based on the timestamp of the actual engine speed signal, and performing linear interpolation correction on the data with time-series deviations. Finally, normalization is performed by mapping all signal values ​​to the [0,1] interval using a normalization formula according to the measurement range of each signal. The preprocessed high-quality data is then stored in the algorithm input area (specified RAM address) of the engine electronic control unit for subsequent use by algorithm modules such as operating condition matching and dynamic correction.

[0094] 3. Precise working condition matching algorithm:

[0095] To address the engine control oscillation problem caused by incomplete coverage of operating points or sudden parameter changes during operating condition switching in traditional operating condition matching, a four-level collaborative algorithm is used to achieve a precise and smooth transition of basic proportional-integral parameters from the mapping table to the actual engine operating conditions, ensuring the continuity and smoothness of parameters as operating conditions change.

[0096] Through a four-level collaborative algorithm—coarse matching, fine interpolation, boundary correction, and transition smoothing—the four algorithms advance sequentially and cooperate with each other to form a complete operating condition matching logic. The coarse matching algorithm quickly locates the two-dimensional operating condition interval to which the current operating condition belongs, narrowing the parameter search range and improving matching efficiency. This algorithm sets a reasonable speed and load interval span based on the engine dynamic characteristic mutation points. Engine dynamic characteristic mutation points refer to speed or load points where engine torque output, fuel consumption, and other characteristics change significantly (such as idle speed, maximum torque speed, and rated speed). The interval span setting must avoid these mutation points to ensure that the engine dynamic characteristics are consistent within the same interval. The specific logic is as follows: First, obtain the preprocessed real-time target speed and actual load; then read the preset speed interval division table and load interval division table; through interval index matching, determine the speed interval to which the real-time target speed belongs and the load interval to which the actual load belongs. The two are combined to form the two-dimensional operating condition interval of the current operating condition, which corresponds to a 2×2 parameter matrix in the basic proportional-integral parameter mapping table.

[0097] To achieve continuous adaptation between discrete operating points and solve the problem of missing non-calibrated operating parameters caused by the discreteness of operating points in the mapping table, the following approach is used. Since the operating points in the mapping table are discrete points divided by a fixed step size, while the actual operating conditions of the engine are continuously changing, directly calling the parameters of discrete operating points will lead to abrupt parameter changes. The fine interpolation algorithm adopts bilinear interpolation, which obtains parameter values ​​for any continuous point by performing linear interpolation on four discrete points in a two-dimensional plane. This method is characterized by its simplicity and high accuracy. Specifically, the logic is as follows: First, the basic proportional gain and basic integral gain parameters of the four vertices of the current two-dimensional operating interval are extracted from the basic proportional-integral parameter mapping table, denoted as P11, P12, P21, P22 and I11, I12, I21, I22, where the first subscript is the speed interval index, the second... The subscript is the load interval index; then the normalized position α of the real-time target speed within the corresponding speed interval (α = (real-time speed - interval start speed) / interval span) and the normalized position β of the real-time actual load within the corresponding load interval (β = (real-time load - interval start load) / interval span) are calculated; then the preliminary basic proportional gain P and preliminary basic integral gain I of the current operating point are calculated using the bilinear interpolation formula: P = P11×(1-α)×(1-β) + P12×(1-α)×β + P21×α×(1-β) + P22×α×β; I = I11×(1-α)×(1-β) + I12×(1-α)×β + I21×α×(1-β) + I22×α×β; the parameters obtained by this calculation can change continuously with the operating condition, avoiding parameter abrupt changes.

[0098] Boundary correction algorithms prevent parameter anomalies at the boundary operating points of the mapping table. When the engine's actual operating point approaches the boundary of the mapping table (e.g., speed close to rated speed, load close to full load), directly using interpolation algorithms may cause parameters to exceed the engine's safe operating range, such as excessive proportional gain leading to severe torque overshoot. Boundary correction algorithms employ a constrained extrapolation strategy, correcting the initially interpolated parameters by setting a linearly decaying extrapolation coefficient to ensure the parameters always remain within a safe range. Specifically, the logic is as follows: first, set the boundary thresholds of the mapping table; the speed boundary thresholds are 95% of the rated speed (upper boundary) and 105% of the idle speed (lower boundary). The load boundary thresholds are 95% of the maximum load (upper boundary) and 5% of the 0% load (lower boundary). The system determines whether the real-time operating point is within the boundary region, such as if the speed is greater than 95% of the rated speed. If it is within the boundary region, the system calculates the distance d from the operating point to the boundary (e.g., d = real-time speed - rated speed × 95%) and the total width D of the boundary region (e.g., D = rated speed - rated speed × 95%). The extrapolation coefficient k = 1 - d / D (k ranges from 0 to 1). The parameters obtained from the initial interpolation are multiplied by the extrapolation coefficient k to obtain the corrected parameters, ensuring that the parameters decrease linearly with distance from the boundary to avoid exceeding the safe range.

[0099] The transition smoothing algorithm ensures parameter stability during engine operating condition switching. When the engine rapidly switches from one operating condition to another (e.g., idle speed → partial load, partial load → full load), the rate of change of speed and load is large. Directly using interpolated parameters can lead to torque output fluctuations and control oscillations. The transition smoothing algorithm uses a first-order low-pass filter to smooth the parameters, suppressing abrupt changes and ensuring a smooth transition. The specific logic is as follows: First, set the filter time constant τ (determined based on the engine's dynamic response characteristics, typically 0.1-0.5s), and the filter coefficient α = control period T / (T + τ). (T=10ms); To determine whether the engine is in the process of switching operating conditions, the engine speed change rate and load change rate are calculated. If the change rate is greater than the preset threshold (speed change rate > 50r / (min・cycle), load change rate > 10% maximum load / cycle), then the engine is in the process of switching operating conditions. If the engine is in the process of switching operating conditions, a first-order low-pass filter formula is used to smooth the parameters after preliminary interpolation or boundary correction: smoothed parameter = α × current cycle preliminary parameter + (1-α) × previous cycle smoothed parameter. Through this filtering process, the parameters can slowly follow the changes in operating conditions, avoid abrupt changes, and ensure stable torque output.

[0100] In practice, the algorithm is first initialized by loading the basic proportional-integral parameter mapping table into the RAM of the engine electronic control unit, setting the interval division rules for coarse matching, the calculation parameters for fine interpolation, the threshold and extrapolation coefficient rules for boundary correction, the filter time constant and switching judgment threshold for transition smoothing, and initializing the historical parameters of the filtering algorithm, i.e., the parameters after smoothing in the previous cycle. Next, coarse matching is performed: the preprocessed real-time target speed and actual load are retrieved, and the two-dimensional operating condition interval to which the current operating condition belongs is determined through interval index matching. The basic proportional-integral parameters of the four vertices of this interval are extracted. Finally, fine interpolation is performed: the real-time speed and load within the assigned interval are calculated. The system first normalizes the position and calculates the initial base proportional gain and initial base integral gain for the current operating point using a bilinear interpolation formula. Then, it performs boundary correction: determining whether the real-time operating point is within the boundary region of the mapping table. If it is, it calculates extrapolation coefficients to correct the initial interpolation parameters. Next, it performs transition smoothing: calculating the rate of change of speed and the rate of change of load, and determining whether the system is in a switching process. If it is, it uses a first-order low-pass filter to smooth the parameters. Finally, it outputs the final base proportional-integral parameters after boundary correction and transition smoothing, and stores them in a designated RAM address of the engine electronic control unit for subsequent dynamic correction module calls.

[0101] 4. Speed ​​control deviation calculation and operating condition identification:

[0102] This invention addresses the problem of lagging correction strategies caused by the inability of traditional single deviation signals to distinguish between transient and steady-state conditions. Traditional correction strategies adjust parameters based solely on the magnitude of the deviation, which can lead to contradictions such as slow transient response and large steady-state overshoot. By accurately calculating speed control deviation, quantifying deviation change trends, and identifying operating conditions, this invention provides the dynamic correction module with precise engine control status information, ensuring accurate matching between the correction strategy and the control status.

[0103] The integrated processing logic encompasses speed control deviation calculation, sliding average preprocessing, speed deviation change rate calculation, and precise operating condition identification. Each step progresses progressively, from raw signals to precise state identification, providing comprehensive and accurate input data for dynamic correction. Speed ​​control deviation calculation is the foundation of the entire technical solution, directly reflecting the degree of deviation in current engine speed control and serving as the core basis for adjusting correction parameters. The speed control deviation is defined as the real-time difference between the target speed and the actual speed, i.e., speed control deviation e = target speed - actual speed. When e > 0, it indicates that the actual speed is lower than the target speed, requiring an increase in the proportional-integral parameter to improve torque and accelerate speed increase. When e < 0, it indicates that the actual speed is higher than the target speed, requiring a decrease in the proportional-integral parameter to reduce torque and slow speed increase. When e = 0, it indicates that the speed has reached the target value, and no parameter adjustment is needed. To ensure the accuracy of deviation calculation, both the target speed and the actual speed use preprocessed signals to avoid deviation calculation errors caused by noise in the raw signals.

[0104] Moving average preprocessing is crucial for improving the stability of deviation signals. Since the original deviation signal may experience instantaneous fluctuations due to external interference, directly using the original deviation for subsequent calculations can lead to incorrect adjustments in the correction strategy, such as excessive parameter adjustment due to increased instantaneous deviation. Moving average preprocessing, by averaging the deviation values ​​over multiple consecutive control cycles, effectively suppresses instantaneous fluctuations and yields a more stable filtered deviation. The specific logic is as follows: The moving average window size N is set based on the engine's control cycle and interference characteristics, typically 5-10 control cycles. The original deviation values ​​e1, e2, ..., eN from the most recent N control cycles are stored. The filtered deviation is e_filter = (e1 + e2 + ... + eN) / N. This processing smooths the fluctuations in the deviation signal, preserves the overall trend of deviation change, and provides a stable input for subsequent deviation change rate calculations and operating condition identification.

[0105] Calculating the rate of change of engine speed deviation is the core of quantifying the trend of deviation change. The control state of the engine cannot be determined solely by the magnitude of the deviation. For example, the same deviation e=100r / min could represent a transient state of rapid speed increase or a steady state of slow speed increase. The rate of change of deviation accurately reflects the speed and direction of deviation change, providing crucial information for identifying operating conditions. The rate of change of engine speed deviation is defined as the ratio of the difference between the filtered deviation of the current control cycle and the previous control cycle to the engine control cycle, i.e., the rate of change of deviation ec=(e_filter_current-e_filter_prev) / T, where e_filter_current is the filtered deviation of the current cycle, e_filter_prev is the filtered deviation of the previous cycle, and T is the control cycle (10ms). When ec>0, it indicates that the deviation is increasing (e.g., the actual speed is lower than the target speed and the gap is widening), requiring increased correction. When ec<0, it indicates that the deviation is decreasing (e.g., the actual speed is lower than the target speed and the gap is narrowing), allowing for reduced correction. When ec=0, it indicates that the deviation is stable and the current correction strategy can be maintained.

[0106] Operating condition identification uses clearly defined thresholds based on the rate of change of engine speed deviation and the absolute value of the deviation after filtering. The engine's operating conditions are categorized into three types: transient, steady-state, and transient processes. Each type corresponds to a different dynamic correction strategy to ensure precise matching between the correction strategy and the control state. The specific classification logic is as follows: First, thresholds are set for the three types of operating conditions. The threshold for a transient process is a deviation change rate > 50 r / (min·cycle) and a filtered absolute deviation value > 50 r / min; the threshold for a steady-state process is a deviation change rate < 10 r / (min·cycle) and a filtered absolute deviation value < 10 r / min; the transient process is an operating condition between transient and steady-state, not meeting either the transient or steady-state thresholds. The transient process corresponds to scenarios where engine speed changes rapidly. For example, during cold starts or speed adjustments after sudden load changes, a rapid response to deviation changes is required, employing a correction strategy with high proportional gain and low integral gain. Steady-state processes correspond to scenarios where engine speed is stable, such as warm engine idling or constant load operation. In these cases, speed stability must be ensured, using a correction strategy with low proportional gain and high integral gain to eliminate steady-state deviations. Transient processes correspond to scenarios with moderate speed changes, requiring a balance between response speed and stability. A correction strategy with medium proportional gain and medium integral gain is used to achieve a smooth transition from transient to steady state. Simultaneously, to ensure the stability of operating condition recognition, a hysteresis threshold is set to avoid frequent switching between different operating conditions. A state switch is only determined when the operating condition meets the threshold of the new state and persists for three control cycles.

[0107] In specific implementation, the algorithm initialization is first completed: the moving average window size is set to N = 5 control cycles, the operating condition judgment threshold is set (transient: ec > 50 r / (min・cycle) and |e_filter| > 50 r / min; steady state: ec < 10 r / (min・cycle) and |e_filter| < 10 r / min), the hysteresis threshold is 3 control cycles, and the historical deviation data (deviation values ​​of the previous N cycles) and the filtered deviation e_filter_prev of the previous cycle are initialized in the moving average window; then the speed control deviation is calculated: the preprocessed target speed and actual speed of the current cycle are retrieved, and the original deviation e = target speed - actual speed is calculated; then the moving average preprocessing is performed: the original deviation e of the current cycle is stored in the moving average window, the oldest deviation value in the window is deleted, the average of the N deviation values ​​in the window is calculated, and the filtered deviation e_filter of the current cycle is obtained; then the speed deviation change rate is calculated: the previous cycle's target speed and actual speed are retrieved, and the filtered deviation e_filter of the previous cycle is calculated; then the speed deviation change rate is calculated: the previous cycle's target speed and actual speed are retrieved, and the filtered deviation e_filter of the previous cycle is calculated; then the historical deviation data (deviation values ​​of the previous N cycles) and the filtered deviation e_filter of the previous cycle are retrieved. The filtered deviation e_filter_prev for one cycle is used to calculate the deviation change rate ec = (e_filter - e_filter_prev) / T (T = 10ms). e_filter_prev is then updated to the current cycle's e_filter for the next cycle's calculation. Next, operating condition identification is performed: the current cycle's ec and |e_filter| are compared with preset operating condition thresholds to determine if transient, steady-state, or transitional conditions are met. If a certain operating condition condition is met, the duration of the current state is recorded. When the duration reaches the hysteresis threshold of 3 cycles, the operating condition is officially determined. If the hysteresis threshold is not reached, the operating condition of the previous cycle is maintained. Finally, the filtered speed control deviation e_filter, speed deviation change rate ec, and corresponding operating condition parameters (transient / steady / transitional) for the current cycle are output to the dynamic correction module and stored in the designated RAM address of the engine electronic control unit for the dynamic correction module to access.

[0108] 5. Dynamic collaborative correction mechanism:

[0109] To address the control performance imbalance caused by a single correction strategy under different engine types and operating conditions, this paper proposes a method that achieves precise adaptation of basic proportional-integral parameters to the real-time control state and inherent characteristics of the engine through dual-coefficient collaborative correction, dual-variable coupled value selection, and differentiated calibration.

[0110] The dual-coefficient collaborative correction system, the dual-variable coupled value-taking logic, and the integrated correction logic for differentiated calibration of spark-ignition and compression-ignition engines all work together to ensure that the corrected parameters have real-time performance, adaptability, and stability. The dual-coefficient collaborative correction system is the core element, setting independent proportional gain correction coefficients and integral gain correction coefficients. These two coefficients dynamically correct the basic proportional gain and basic integral gain of the output, respectively, to achieve independent dynamic adjustment of the two types of parameters. At the same time, their synergistic effect ensures that the response speed and stability of engine torque control reach the optimal balance. The core function of the proportional gain correction coefficient kP is to adjust the torque response speed. When kP increases, the torque response speed to speed deviation increases, which can quickly reduce the deviation, but too large a kP can easily lead to overshoot. When kP decreases, the response speed slows down, but the stability improves. The core function of the integral gain correction coefficient kI is to adjust the steady-state deviation elimination capability. When kI increases, the integral action is enhanced, which can quickly eliminate steady-state deviation, but if it is too large, it will easily lead to integral saturation and overshoot. When kI decreases, the integral action is weakened, the steady-state deviation elimination speed slows down, but the stability improves. The core logic of the dual-coefficient coordinated correction is to dynamically adjust kP and kI according to the real-time control status of the engine (operating condition, deviation magnitude, deviation change rate), so that kP and kI cooperate with each other. In the transient process, priority is given to ensuring the response speed (large kP, small kI), in the steady-state process, priority is given to ensuring the steady-state accuracy (small kP, large kI), and in the transient process, the response and stability are balanced (medium kP, medium kI).

[0111] The bivariate coupling value-taking logic is the core basis for the dynamic adjustment of the correction coefficient. Based on the coupling relationship between the absolute value of the deviation |e_filter| and the rate of change of the deviation ec, a clear rule for determining the correction coefficient is established to ensure precise matching between the correction coefficient and the real-time control state of the engine. The core idea of ​​this logic is: the absolute value of the deviation reflects the severity of the current control deviation, and the rate of change of the deviation reflects the trend of the deviation change; both jointly determine the value of the correction coefficient. Specifically, |e_filter| is divided into three intervals (small deviation: |e_filter| < 20 r / min; medium deviation: 20 r / min ≤ |e_filter| ≤ 100 r / min; large deviation: |e_filter| > 100 r / min), and ec is divided into three intervals (small rate of change: ec < 20 r / (min·cycle); medium rate of change: 20 r / (min·cycle) ≤ ec ≤ 80 r / (min·cycle); large rate of change: ec > 80 r / (min·cycle)), forming a 3×3 correction coefficient value matrix. For the proportional gain correction coefficient kP: The maximum value (kP_max=1.5) is used for the large deviation and high rate of change range; a relatively large value (kP=1.3) is used for the large deviation and medium rate of change range; a relatively large value (kP=1.2) is used for the medium deviation and medium rate of change range; the median value (kP=1.0) is used for the medium deviation and medium rate of change range; and the minimum value (kP_min=0.5) is used for the small deviation and small rate of change range. For other ranges, kP values ​​are determined by linear interpolation. The kP value increases with the increase of |e_filter| and ec to ensure a fast response when the deviation is large and the change is rapid. For the integral gain correction coefficient kI: large deviation... The minimum value (kI_min=0.3) is taken for the large deviation and large rate of change interval, the smaller value (kI=0.5) is taken for the large deviation and medium rate of change interval, the smaller value (kI=0.6) is taken for the medium deviation and large rate of change interval, the middle value (kI=1.0) is taken for the medium deviation and medium rate of change interval, and the maximum value (kI_max=1.8) is taken for the small deviation and small rate of change interval. The kI value is determined by linear interpolation for other intervals. The kI value decreases as |e_filter| and ec increase to ensure that overshoot is suppressed when the deviation is large and the change is rapid, and the integral effect is enhanced to eliminate steady-state deviation when the deviation is small.

[0112] Differential calibration is necessary because spark-ignition engines (such as gasoline engines) and compression-ignition engines (such as diesel engines) have significantly different combustion characteristics, resulting in different dynamic response characteristics. Therefore, it is necessary to preset the value range of correction coefficients for each type of engine to ensure that the correction strategy accurately adapts to the inherent characteristics of different engines. Spark-ignition engines have a fast combustion speed (flame propagation speed 20-30m / s), a sensitive engine response, and torque output changes rapidly with parameter adjustments, making them prone to overshoot. Therefore, the value range of its correction coefficients needs to focus on stability, with kP set to 0. The value of kI ranges from 0.5 to 1.6. Compression ignition engines have slow combustion speeds (flame propagation speed 5-10 m / s), resulting in engine response lag and slow torque output changes with parameter adjustments, making them prone to response lag. Therefore, the correction coefficient range should focus on response speed, with kP ranging from 0.8 to 1.5 and kI ranging from 0.3 to 1.8. At the same time, for different displacements and cylinder numbers of the same type of engine, targeted fine-tuning calibration is also required to ensure that the correction strategy is adapted to the specific characteristics of the engine.

[0113] In specific implementation, the algorithm initialization is first completed: Based on the engine type (spark ignition / compression ignition), the corresponding correction coefficient range is loaded, and the three-interval division of |e_filter| (small deviation < 20 r / min, medium deviation 20 r / min - 100 r / min, large deviation > 100 r / min) and the three-interval division of ec (small rate of change < 20 r / (min·cycle), medium rate of change 20 r / (min·cycle) - 80 r / (min·cycle), large rate of change > 80 r / (min·cycle)) are set. A 3×3 kP and k... The I-value matrix is ​​initialized, along with the correction coefficients and corrected parameters from the previous cycle. Next, the core input parameters are retrieved: the base proportional gain P_base, base integral gain I_base, filtered deviation |e_filter|, deviation change rate ec, and current operating condition are read from the specified address of the engine electronic control unit. Then, engine type determination is performed: based on pre-stored engine structural parameters, it is determined to be either a spark-ignition or compression-ignition engine, and the corresponding upper and lower limits for kP and kI values ​​are determined. Finally, bivariate coupling is performed: |e_filter| and ec are matched to the corresponding 3×3 matrix intervals. The initial correction coefficients for the given interval are read. If the interval is within the boundary, precise kP and kI values ​​are determined through linear interpolation to ensure continuous adjustment of the correction coefficients with deviation and rate of change. Simultaneously, a secondary calibration is performed based on the current operating conditions. During transient processes, kP is increased by 10%-15% and kI is decreased by 10%-15%. During steady-state processes, kP is decreased by 10%-15% and kI is increased by 10%-15%. The matrix values ​​remain unchanged during the transient process to avoid correction deviations caused by single variable values. Subsequently, a dual-coefficient collaborative correction is performed: the base proportional gain is multiplied by the calibrated kP to obtain the corrected proportional gain P_adj = P_ba. se×kP; Multiply the base integral gain by the calibrated kI to obtain the corrected integral gain I_adj=I_base×kI; During the correction process, ensure that P_adj and I_adj are always within the range corresponding to the engine type. If they exceed the range, take the boundary value and mark the calibration status; Finally, output the corrected proportional-integral parameters: Store P_adj and I_adj in the specified RAM address of the engine electronic control unit, and record the corrected kP, kI, |e_filter|, ec, and other data for subsequent mapping table iteration optimization and fault diagnosis, thus completing this dynamic collaborative correction process.

[0114] 6. Optimal proportional-integral parameter synthesis and security verification:

[0115] To address engine control malfunctions caused by parameters exceeding range, sudden changes, or integral saturation, an integrated logic combining coupled synthesis and multi-layer safety verification is used to ensure that the parameters output to the torque control system are safe, reliable, and real-time, while avoiding problems such as speed runaway and torque surges caused by abnormal parameters.

[0116] The integrated processing logic, consisting of basic parameter-correction coefficient coupling synthesis, multi-layer safety verification mechanism, and closed-loop execution of the control cycle, forms a safety barrier for parameter output. The coupling synthesis algorithm, the foundation for generating preliminary optimal parameters, employs the logic of directly multiplying the basic proportional-integral parameters with the dynamic correction coefficients. This algorithm is simple, efficient, and has short computation time, quickly generating preliminary optimal proportional gain P_init and preliminary optimal integral gain I_init to meet the real-time control requirements of the engine. The synthesis logic and dynamic correction results are seamlessly integrated, ensuring parameter continuity. The multi-layer safety verification mechanism is the core of this technical solution, comprising three core components: safety range verification, parameter change rate limitation, and special operating condition rationality verification. These three components are executed sequentially, with each layer providing checks: the safety range verification is based on the engine's structural characteristics and operational safety requirements, presetting absolute safety thresholds for the proportional and integral gains. These thresholds are determined through extensive bench testing. The system is designed to cover the safe operating range under all working conditions, preventing abnormal torque output caused by parameters exceeding thresholds. Parameter change rate limiting is used to suppress sudden and large changes in parameters, setting the maximum allowable change rate of parameters within a single control cycle to avoid torque shocks caused by parameter mutations and ensure smooth engine power output. Special working condition rationality verification targets severely deviated working conditions to prevent integral saturation caused by excessive integral gain. Integral saturation can lead to engine speed overshoot and torque runaway. This step forcibly reduces the integral gain and pauses integral accumulation until the deviation returns to a reasonable range, thus preventing integral saturation at its source. The closed-loop execution mechanism of the control cycle is crucial for ensuring real-time performance. Within each engine control cycle, the entire process of coupled synthesis and multi-layer safety verification is fully executed, ensuring that the total delay from signal acquisition to parameter output does not exceed one control cycle. Simultaneously, the optimal parameters after verification are fed back to the preceding modules in real time, forming a closed-loop parameter management system.

[0117] In specific implementation, the algorithm initialization is first completed: the preset proportional gain safety threshold (P_min-P_max) and integral gain safety threshold (I_min-I_max) are loaded, the maximum rate of change of P in a single control cycle ΔP_max and the maximum rate of change of I ΔI_max are set, the severe deviation judgment threshold (|e_filter|≥300r / min) and the lower limit of I_sat for integral saturation suppression are determined, and the optimal proportional gain P_prev and optimal integral gain I_prev of the previous cycle are initialized; then the coupled synthesis input parameters are retrieved: the corrected proportional gain P_adj and corrected integral gain I_adj, and the filtered deviation |e_filter| are read from the specified address of the engine electronic control unit; then the coupled synthesis is performed: P_adj is used as the initial optimal proportional gain P_init, and I_adj is used as the initial optimal integral gain I_init, completing the generation of the initial optimal parameters.

[0118] Then, multi-layer safety checks are performed: The first step is a safety range check, comparing P_init with P_min - P_max. If P_init < P_min, it is replaced with P_min; if P_init > P_max, it is replaced with P_max, resulting in the range-checked parameter P_range. Similarly, I_init is compared with I_min - I_max to obtain I_range. If parameter replacement occurs, an out-of-range status is marked for fault diagnosis. The second step is parameter change rate limiting, calculating the rate of change ΔP = |P_range - P_prev| / If ΔP > ΔP_max, then P_range is adjusted to P_prev + ΔP_max × T to obtain the rate-limited parameter P_rate. Similarly, ΔI is calculated and I is adjusted to obtain I_rate, ensuring that the parameter changes smoothly. The third step is to perform a special operating condition rationality check, to determine whether |e_filter| is ≥ the severe deviation threshold. If it is satisfied, it is determined to be a severe deviation operating condition, and I_rate is forcibly adjusted to I_sat to obtain the verified integral gain I_check. At the same time, the integral saturation suppression state is marked. If it is not satisfied, I_rate is directly used as I_check, and P_rate is directly used as P_check.

[0119] Finally, closed-loop output and parameter update are performed: P_check and I_check are determined as the final optimal proportional-integral parameters, stored in the designated address of the engine electronic control unit, and output to the engine torque control system; at the same time, P_check is updated to P_prev and I_check is updated to I_prev for the calculation of the parameter change rate in the next control cycle, thus completing this synthesis and verification process. The entire process is executed in a closed loop within one control cycle to ensure real-time performance.

[0120] 7. Coordinated control of proportional-integral scheduling and feedforward compensation:

[0121] To address the response lag issue of traditional pure proportional-integral feedback control under sudden load changes, a composite control logic combining feedback adjustment and feedforward prediction is used to achieve rapid response and stable output of torque control, significantly reducing speed fluctuations during sudden load changes and improving engine operation stability.

[0122] The integrated collaborative logic of load step recognition, three-dimensional feedforward compensation mapping table, dynamic weight superposition strategy, and composite torque control command generation is seamlessly connected, achieving an organic combination of feedback and feedforward. Load step recognition is the prerequisite for triggering feedforward compensation. By calculating the load change rate in real time and comparing it with a preset threshold, it accurately identifies typical load step scenarios such as air conditioning start-up, power take-off operation switching, remote throttle sudden changes, and sudden load changes in construction machinery, providing a precise basis for triggering feedforward compensation. Moreover, the recognition logic is based on pre-processed load signals, ensuring the accuracy and real-time performance of the recognition. The three-dimensional feedforward compensation mapping table is the core source of feedforward compensation values. It is constructed based on three dimensions: target speed, actual load, and load change amplitude. Through bench calibration, it stores the optimal torque compensation values ​​under different operating conditions. This mapping table is calibrated in the same way as the basic proportional-integral parameter mapping table. To ensure the accuracy and adaptability of the compensation values, covering all typical load step scenarios of the engine; the dynamic weight superposition strategy is the key to balancing the effects of feedback and feedforward. According to the engine's operating state (transient / steady state), the weights of the feedforward compensation value and the proportional-integral feedback torque are dynamically adjusted. In the transient process, the feedforward weight is increased to strengthen the predictive effect and quickly respond to load changes. In the steady state process, the feedforward weight is reduced to weaken the predictive effect and ensure the steady-state accuracy of feedback regulation. In the transient process, the weights are adjusted by linear interpolation to achieve a smooth transition between the effects of the two; the generation of composite torque control commands is the ultimate goal of coordinated control. The feedforward compensation value and the proportional-integral feedback torque are linearly superimposed according to dynamic weights to generate the final torque control command, which is sent to the engine actuator through the bus to achieve composite control of feedback regulation and feedforward prediction, taking into account both dynamic response speed and steady-state control accuracy.

[0123] In specific implementation, the algorithm initialization is first completed: the three-dimensional feedforward compensation mapping table is loaded, the load step judgment threshold is set, the load change rate is ≥10% of the maximum load / control cycle, the transient value (0.33-0.5) and steady-state value (0.17-0.25) of the feedforward weight are set, the load value, feedforward weight, and feedforward compensation value of the previous cycle are initialized, and the bus communication parameters (sending cycle 10ms, message format) are configured; then the core input parameters are retrieved: the final optimal proportional-integral parameters, the preprocessed current load value, the target speed, and the load value of the previous cycle are read from the specified address of the engine electronic control unit; then the load step recognition is performed: the current load change rate = |current load value - previous cycle load value| / control cycle is calculated, the load change amplitude = current load value - previous cycle load value is calculated, if the load change rate ≥ judgment threshold, it is judged as a load step scenario and marked as transient state, if it does not reach the threshold, it is judged as a steady-state scenario.

[0124] Then, the feedforward compensation value is retrieved: if the scenario is determined to be a load step scenario, the corresponding optimal torque compensation value FF_Torque is retrieved from the three-dimensional feedforward compensation mapping table using a three-dimensional interpolation algorithm based on the current target speed, actual load, and load change amplitude; if it is a steady-state scenario, FF_Torque is set to 0; then, dynamic weight calculation is performed: if it is a transient state, the preset transient feedforward weight W_FF is selected; if it is a steady state, the steady-state feedforward weight W_FF is selected; if it is a transient state, W_FF is adjusted by linear interpolation, and the weight of the proportional-integral feedback torque W_PI = 1 - W_FF is determined; then, the proportional-integral feedback torque calculation is performed: the final optimal proportional-integral parameters are input into the proportional-integral control. The actuator, combined with the speed control deviation, calculates the feedback torque PI_Torque; then, it executes the generation of composite torque control commands: the final torque control command Final_Torque = FF_Torque × W_FF + PI_Torque × W_PI is calculated by weighting and superimposing, ensuring that Final_Torque is within the range of the engine's maximum and minimum torque; finally, the command is sent: Final_Torque is encapsulated in a preset bus message format and sent to the engine actuator via the CAN2.0 / CANFD bus at a period of 10ms, while recording data such as load changes, feedforward compensation values, and weight values ​​for subsequent mapping table iteration optimization.

[0125] 8. Iterative optimization of the basic proportional-integral parameter mapping table throughout its entire lifecycle:

[0126] To address the issue of decreased control performance caused by changes in engine characteristics, component wear, and carbon buildup after long-term engine use, a dynamic iteration of mapping table parameters is achieved through a full-process quantitative evaluation, data recording, optimization judgment, and parameter updates, ensuring that the engine maintains good torque control performance throughout its entire service life.

[0127] The integrated optimization logic, comprising four-dimensional quantitative evaluation indicators, data recording categorized by operating point, optimization triggering based on work cycles, update conditions for multi-environment verification, and parameter backtracking and batch calibration mechanisms, forms a complete optimization closed loop, ensuring the accuracy, reliability, and timeliness of optimization. The four-dimensional quantitative evaluation indicators are the foundation for comprehensively evaluating torque control performance. Indicators are set from three core dimensions: control accuracy, response speed, and stability, including torque control deviation rate, response time, overshoot, and oscillation frequency. All four indicators have been calibrated to pass thresholds through bench testing, comprehensively and accurately quantifying the control effect at each operating point, providing an objective basis for optimization judgment. Data recording categorized by operating point is crucial for achieving precise optimization. Using the mapping table operating points as units, the operating data for each operating point is recorded independently, ensuring a one-to-one correspondence between parameters and control effects, avoiding optimization errors caused by data confusion from different operating points. The recorded data includes optimal proportional-integral parameters, control effect indicators, and operating environment parameters, providing complete data support for optimization judgment. The optimization based on work cycles... Optimization triggering is the core of balancing optimization frequency and system load. The optimization trigger cycle is set based on the engine's operating cycle to avoid excessively frequent optimization calculations that could increase the load on the engine's electronic control unit, while ensuring timely capture of changes in engine characteristics and dynamic parameter updates. Multi-environment verification update conditions are crucial for ensuring the reliability of optimized parameters. Parameters at a given operating point must meet the requirement that evaluation indicators for multiple consecutive control cycles are better than the initial calibration value, and must pass real-vehicle verification in multiple environments such as high temperature, low temperature, and high altitude before being updated to the mapping table. This ensures that the updated parameters have good adaptability under different environmental conditions. Parameter backtracking and batch calibration mechanisms are important supplements to optimization reliability. The parameter backtracking mechanism can quickly restore parameters to historical optimal values ​​when abnormal parameter updates cause a decline in control performance, preventing control performance degradation. The batch calibration mechanism can optimize the consistency of parameters at adjacent operating points when the cumulative updated operating points reach a certain proportion, avoiding oscillations during operating point switching caused by excessive differences in local parameters, and ensuring the overall adaptability of the mapping table.

[0128] In practical implementation, the algorithm and parameters are first initialized: four-dimensional quantitative evaluation indicators and qualified thresholds are set, torque control deviation rate ≤2%, response time ≤0.2s, overshoot ≤5%, oscillation frequency ≤1 time / condition switching, the optimization trigger cycle is set to 1000 engine working cycles, multi-environment verification requirements are set, and real vehicle verification is completed at high temperature 45℃±2℃, low temperature -25℃±2℃, and high altitude 85kPa±2kPa, parameter backtracking conditions are set (any evaluation indicator exceeds 1.5 times the qualified threshold), and the batch calibration trigger ratio is set to the total operating points of the mapping table. 30% configuration of data recording storage address and format, initialization of data recording module and optimization judgment module; then execute data recording by operating point classification: during actual engine operation, the operating data of each operating point is recorded in real time, based on the mapping table operating points, including the final optimal proportional integral parameter output, the real-time value of the four-dimensional evaluation index, and the operating environment parameters (ambient temperature, atmospheric pressure). The data recording cycle is synchronized with the engine control cycle, and the data of each operating point is stored independently in the Flash memory of the engine electronic control unit to ensure that the data is not lost.

[0129] Then, the optimization trigger and judgment are executed: when the engine has completed 1000 working cycles, the optimization judgment process is triggered. The recorded data of each working point is statistically analyzed. First, it is determined whether the four-dimensional evaluation index of the working point for 50 consecutive control cycles is better than the initial calibration value. If it is satisfied, multi-environment real vehicle verification is started. On the real vehicle verification platform in a typical environment, the working point is tested in the whole process to verify whether its evaluation index in different environments meets the qualified threshold. If the multi-environment verification is qualified, it is determined that the optimization conditions are met. If any link is not met, the optimization is abandoned. Then, the parameter update is executed: for the working point that meets the optimization conditions, the recorded final optimal proportional-integral parameter is updated to the new calibration value of the basic proportional-integral parameter mapping table. The new parameter is written to the Flash memory of the engine electronic control unit through the algorithm, replacing the original parameter. At the same time, the parameter update time, environment, evaluation index and other information are recorded to form an optimization log.

[0130] Simultaneously, a parameter backtracking mechanism is implemented: after parameter updates, the four-dimensional evaluation indicators of the operating point are continuously monitored. If any indicator exceeds 1.5 times the qualified threshold, it is determined that the parameter update is abnormal, and parameter backtracking is immediately triggered. Within 10 control cycles, the parameters of the operating point are restored to the historical optimal value, and the abnormal log is marked for subsequent fault analysis and re-optimization. Finally, batch calibration is performed: when the cumulative number of updated operating points reaches 30% of the total operating points in the mapping table, the batch calibration process is triggered. Clustering algorithms are used to analyze the parameters of all operating points, optimize the consistency of parameters of adjacent operating points, eliminate the problem of excessive local parameter differences, ensure smooth parameter transition when switching operating conditions, and batch write the calibrated parameters to the mapping table to complete this batch optimization. The entire iterative optimization process is automatically executed during engine operation without manual intervention.

[0131] 9. Execution and Adaptation:

[0132] This invention addresses the challenges of widespread adoption of traditional engine control technologies due to their strong hardware dependence, poor adaptability, and weak fault tolerance. By employing an integrated implementation logic that enables full-process execution of existing engine electronic control units, emergency handling of typical faults, wide-range engine adaptation, and pure software implementation, the method can be rapidly and cost-effectively implemented on existing engine systems, while also possessing broad adaptability and high reliability.

[0133] The core of this technical solution lies in the full-process execution of existing engine electronic control units (ECUs), emergency handling of typical faults, wide-range engine adaptation, and pure software implementation. These core components support each other, balancing ease of implementation, operational reliability, and versatility for widespread adoption. The full-process execution of existing engine ECUs is key to reducing hardware costs. Using the existing engine ECU as the execution carrier for all technical solutions eliminates the need for any additional hardware, fully utilizing existing hardware resources and significantly reducing engine modification costs and industrialization barriers. Furthermore, existing engine ECUs possess full-process capabilities including data storage, signal acquisition, parameter calculation, and bus communication, fully meeting the computational and execution requirements of this invention. Emergency handling of typical faults is crucial for improving operational reliability. It pre-defines typical fault scenarios during engine operation, such as abnormal sensor data, parameter calculation failures, and bus communication failures, and pre-sets corresponding solutions for each type of fault. The emergency proportional-integral parameter, obtained through full-condition statistics, ensures the engine maintains basic operation under fault conditions. A rapid fault detection and parameter switching mechanism is also established, ensuring emergency parameter switching is completed within 10ms after fault detection, preventing engine shutdown due to faults and improving operational reliability. Wide-range engine adaptability is the core of this invention's versatility. The method is compatible with 1.0L-15.0L displacement, 2-12 cylinder spark-ignition and compression-ignition engines, and adapts to the control systems of mainstream engine brands such as Weichai, Yuchai, and Cummins. Only specific parameter calibration is required for different engine characteristics, without modifying the core algorithm, significantly expanding the invention's applicability and reducing the difficulty of industrialization. Pure software implementation is key to simplifying the implementation process. No modification to the existing engine hardware structure is required; only the software algorithm of the engine electronic control unit is ported, integrated, and calibrated.

[0134] In specific implementation, the compatibility verification of existing engine electronic control units is first completed: for engine electronic control units of different brands and types, it is verified whether their hardware resources (storage capacity, computing power, signal interface, bus interface) meet the execution requirements of the method of this invention. If the hardware resources meet the requirements, they are used directly; if the storage capacity is insufficient, the Flash memory is expanded to ensure that the algorithm code, mapping table and running data can be stored. Then, the algorithm is ported and integrated: the algorithm code of all technical solutions of this invention (including signal acquisition preprocessing, precise matching of working conditions, dynamic collaborative correction, optimal parameter synthesis verification, feedforward compensation collaborative control, mapping table iterative optimization, fault tolerance, etc.) is converted into the code format supported by the engine electronic control unit through the porting tool. The algorithm module is integrated with the existing control program of the engine electronic control unit, and the docking of each algorithm module with the hardware interfaces such as signal acquisition, bus communication, and command output is completed to ensure smooth data interaction without conflict or lag.

[0135] Then, targeted parameter calibration is performed: for different types and displacements of engines, the core parameters such as the basic proportional-integral parameter mapping table, the three-dimensional feedforward compensation mapping table, the dynamic correction coefficient, the safety verification threshold, and the emergency proportional-integral parameters are calibrated through bench testing. The calibration process follows the calibration rules of the aforementioned technical solution of this invention to ensure that the parameters are adapted to the characteristics of the specific engine, and the calibration process is simplified compared to the traditional fixed parameter calibration. Subsequently, a fault emergency handling strategy is configured: typical fault scenarios such as sensor data abnormality (signal exceeds the normal range by 5% for 3 control cycles), parameter calculation failure (calculation result is empty or exceeds the safety threshold), and bus communication failure (bus interruption for 2 control cycles) are preset. Corresponding emergency proportional-integral parameters are configured for each type of fault, and fault detection logic and emergency parameter switching mechanism are set to ensure that parameter switching is completed within 10ms after a fault is detected. At the same time, fault indicator light illumination and fault code storage functions are configured. The fault code contains information such as fault type, occurrence time, and fault status for later fault diagnosis.

[0136] Next, a wide-range real-vehicle adaptation verification was performed: the engine electronic control unit integrating the algorithm of this invention was installed on spark-ignition and compression-ignition engines of different displacements and cylinder numbers. Real-vehicle verification was carried out in various application scenarios such as vehicle driving, remote throttle operation, power take-off operation, and construction machinery operation. The torque control performance, dynamic response performance, and fault tolerance capability under all working conditions were tested to verify that the method of this invention can be adapted to various application scenarios and that the control performance meets the requirements. Finally, pure software implementation and mass production were completed: the debugged and calibrated algorithm and parameters were solidified into the Flash memory of the engine electronic control unit to form a standardized software program. In the mass production process, it is only necessary to flash this software program to the engine electronic control unit and complete the targeted parameter fine-tuning to realize the mass implementation of the method of this invention, and no modification to the engine hardware structure is required throughout the process.

[0137] Finally, it should be noted that the above embodiments are merely examples for clearly illustrating the present invention and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A PI parameter scheduling method for improving the adaptability of engine torque control, characterized in that, The scheduling method involves the following steps: A two-dimensional, hierarchical, decoupled adaptive gain scheduling system is constructed, consisting of an operating condition dimension and a deviation dimension. This system adapts to the dynamic characteristics of the engine body through the operating condition dimension and adapts to the real-time control state through the deviation dimension. Establish a basic proportional-integral parameter mapping table, which takes the engine target speed and actual load as the core dimensions and covers all engine operating conditions. Collect multi-dimensional engine operating signals, including target speed, actual speed, actual load, and dynamic characteristic signals of operating conditions related to torque control; After preprocessing the collected operating signals, the basic proportional-integral parameters corresponding to the current operating condition are obtained through the operating condition matching algorithm; Calculate the engine speed control deviation and the rate of change of deviation, and construct a dynamic collaborative correction mechanism based on the two to correct the basic proportional-integral parameters; The basic proportional-integral parameters are coupled with the corresponding correction coefficients to calculate the optimal proportional-integral parameters and perform a safety check. The optimal proportional-integral parameters that have passed the verification are input into the engine torque control system to generate torque control commands and send them to the actuators. Continuously monitor the torque control effect and perform full lifecycle iterative optimization of the basic proportional-integral parameter mapping table; An integrated fault-tolerance mechanism is implemented to execute preset emergency handling strategies when a fault is detected, ensuring basic engine operation.

2. The method as described in claim 1, characterized in that, The specific steps for establishing the basic proportional-integral parameter mapping table are as follows: The mapping table structure is designed, including independent proportional gain basic mapping table and integral gain basic mapping table, both of which adopt a two-dimensional adaptation structure of target speed and actual load. Set the target speed range to include the entire operating range of the engine from idle to maximum rated speed; Choose the quantitative representation method of the actual load. You can choose any one of fuel injection quantity, intake air quantity or percentage load, and switch between different representation methods through calibration. The systemic step response test under all operating conditions and in all environments was carried out on the engine bench, in two stages: cold engine and hot engine. Multiple rounds of iterative optimization and calibration were performed on operating points with different speed and load combinations to determine the optimal basic parameters corresponding to each operating point; the parameters were compensated and corrected in combination with environmental factors, and the mapping table parameters were matched with the dynamic characteristics of the engine under various steady-state and transitional operating conditions.

3. The method as described in claim 1, characterized in that, The specific steps for collecting multi-dimensional operating signals of the engine are as follows: The types of signals collected include target speed, actual speed, actual load, and dynamic characteristic signals of operating conditions. Among them, the dynamic characteristic signals of operating conditions cover temperature, pressure, load change rate, and throttle opening. High-precision sensors adapted to various signal detection requirements are selected to form a sensor group; and various signals are synchronously collected through the sensor group at a frequency consistent with the engine control cycle. The collected data are processed sequentially, including outlier handling, noise reduction, synchronization alignment, and normalization preprocessing.

4. The method as described in claim 2, characterized in that, The specific steps for obtaining the basic proportional-integral parameters through operating condition matching are as follows: Based on the real-time target speed and actual load, the engine is positioned to a preset operating range, which is divided according to the engine's dynamic characteristic change point. The parameters of the mapping table within the working condition interval are calculated using an interpolation algorithm to obtain preliminary basic proportional-integral parameters; If the current operating point is close to the boundary of the mapping table, the boundary constraint correction strategy is used to adjust the initial basic parameters; if the engine is in the process of switching operating conditions, the transition smoothing processing strategy is used to optimize the initial basic parameters to control the torque output to be stable and without oscillation.

5. The method as described in claim 1, characterized in that, The specific steps for calculating the speed control deviation and the rate of change of deviation and constructing a dynamic correction mechanism are as follows: Calculate the real-time difference between the engine target speed and the actual speed to obtain the speed control deviation; The speed control deviation is processed using a preset signal preprocessing method to suppress instantaneous fluctuation interference; the quantified value of the speed deviation change trend within a continuous control cycle is calculated to obtain the speed deviation change rate; The transient and steady-state conditions of the operating conditions are identified by the rate of change of speed deviation; the speed control deviation and the rate of change of deviation are set as dual input variables for dynamic correction, and a dynamic collaborative correction mechanism is constructed.

6. The method as described in claim 5, characterized in that, The specific steps for correcting the basic proportional-integral parameters are as follows: A dynamic correction system is constructed, which includes proportional gain correction coefficient and integral gain correction coefficient; The logic for setting the value of the correction coefficient is set and dynamically adjusted based on the bivariate coupling relationship between the absolute value of the speed deviation and the rate of change of the deviation; Based on the differences in combustion characteristics between spark-ignition and compression-ignition engines, the correction coefficients are calibrated differently. If it is a transient process, the correction coefficient is adjusted to enhance the proportional effect and weaken the integral effect; If the process is in a steady state, the correction coefficient is adjusted to weaken the proportional effect and enhance the integral effect; If the process is in transition, the correction coefficient will be dynamically adjusted according to the linear transition logic. The basic proportional-integral parameters are corrected using the adjusted correction factors.

7. The method as described in claim 1, characterized in that, The specific steps for synthesizing the optimal proportional-integral parameters and performing security checks are as follows: The basic proportional-integral parameters are coupled with the corresponding correction coefficients to obtain the preliminary optimal proportional-integral parameters; Perform a safety range check on the preliminary optimal proportional-integral parameters, eliminate parameters that exceed the preset safety boundary, and limit the rate of parameter change; The above coupling and verification process is executed in a closed loop within each engine control cycle, and the control parameter response is synchronized with the changes in operating conditions in real time.

8. The method as described in claim 1, characterized in that, The specific steps for integrated proportional-integral scheduling and feedforward compensation coordinated control are as follows: By monitoring the engine operating status through the rate of load change, scenarios of sudden load changes can be identified. The preset multi-dimensional feedforward compensation mapping table is invoked, and the appropriate torque compensation value is retrieved based on the current target speed, actual load, and load change range. A dynamic weighting strategy is set up to increase the weight of the feedforward compensation value in the transient process and decrease the weight of the feedforward compensation value in the steady-state process. The feedforward compensation value is superimposed with the torque correction command output by the proportional-integral controller according to the set weight to generate the final torque control command and send it to the actuator.

9. The method as described in claim 1, characterized in that, The specific steps for iteratively optimizing the basic proportional-integral parameter mapping table are as follows: Set core evaluation indicators, including torque control deviation rate, response time, overshoot, and number of oscillations; During actual engine operation, parameter application data and corresponding control effect data are continuously recorded according to operating point classification. The optimization judgment is triggered according to a preset cycle. If the control effect of a certain working point is consistently better than the initial calibration value and has been verified as qualified by real vehicles in multiple environments, the mapping table parameters of that working point will be automatically updated. Set up a parameter backtracking mechanism so that if the updated parameters cause a decrease in control performance, the system will automatically backtrack to the historical optimal value. When the cumulative updated operating point reaches the preset ratio, batch calibration is initiated to adjust the correlation of parameters of adjacent operating points.

10. The method according to any one of claims 1 to 9, characterized in that, The execution and adaptation steps of this method are as follows: The engine electronic control unit serves as the execution carrier. This electronic control unit integrates a fault-tolerant function module. The fault-tolerant module has a preset emergency handling strategy for scenarios such as abnormal sensor data, parameter calculation failure, and communication failure. When a fault is detected, it switches to emergency proportional-integral parameters within 10ms to control the engine to maintain basic operation. It is compatible with spark-ignition and compression-ignition engines of various displacements and cylinder numbers.

Images