A water pressure dynamic control method for an intelligent watering cart based on MPC prediction

Abstract

This invention provides a dynamic water pressure control method for intelligent sprinkler trucks based on MPC prediction, relating to the fields of wireless communication and intelligent control. The method includes the following steps: S1, data acquisition and preprocessing; S2, pedestrian detection and tracking; S3, generating an optimal avoidance trajectory based on MPC; S4, constructing a multi-condition adaptive water pressure model to regulate the water pressure value; S5, using a PID algorithm for closed-loop water pressure control; S6, switching the sprinkler truck's operating mode according to actual road conditions. Beneficial effects: The method employs multi-sensor perception combined with deep learning and integrated tracking algorithms to achieve accurate acquisition of pedestrian dynamic information; the trajectory planning prediction mechanism using model predictive control upgrades passive response to proactive prevention, reducing safety hazards caused by operational errors; the water pressure model is constructed, and closed-loop control enables real-time adjustment; adjusting the optimization coefficients according to the scenario improves the method's applicability to different scenarios.

Description

Technical Field

[0001] This invention relates to the field of wireless communication and intelligent control, specifically to a method for dynamic water pressure control of an intelligent sprinkler truck based on MPC prediction. Background Technology

[0002] While some existing technologies address vehicle-pedestrian avoidance, employing vision-based active pedestrian avoidance and water pressure adaptive control by installing vision controllers, control room equipment, and water pump control devices on sprinkler trucks, the accuracy of pedestrian detection is affected by single sensors in complex environments such as fog, inclement weather, and tunnels. Furthermore, current technologies struggle to accurately determine the movement state and intentions of pedestrians in complex scenarios. Regarding water pressure control, some automated solutions based on simple control algorithms exist, but they lack comprehensive consideration of multiple factors and intelligent adjustment, failing to achieve precise adaptive water pressure control. Summary of the Invention

[0003] The main objective of this invention is to provide a method for dynamic water pressure control of intelligent sprinkler trucks based on MPC prediction, thereby solving the problems mentioned in the background art.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for dynamic water pressure control of intelligent sprinkler trucks based on MPC prediction, comprising the following steps: S1. Perform data acquisition and preprocessing operations; S2. Perform pedestrian detection and tracking operations; S3. Generate the optimal avoidance trajectory based on MPC; S4. Construct a water pressure adaptive model based on multiple working conditions to regulate the water pressure value; S5. Use the PID algorithm to perform closed-loop control of water pressure; S6. Switch the working mode of the sprinkler truck according to the actual road conditions.

[0005] Furthermore, the data includes: images of the vehicle's surroundings, distance, speed, and angle information of the target object, and the sprinkler truck's own position, heading angle, speed, and acceleration.

[0006] Furthermore, in step S1, millimeter-wave radar data is used to correct the visual detection results. The correction process is as follows: S101. Transform the coordinate system of the millimeter-wave radar and extract pedestrian bounding box information; convert the target position output by the millimeter-wave radar from polar coordinates to Cartesian coordinates; S102. Perform target matching and determination. The specific process is as follows: Calculate the Euclidean distance between the millimeter-wave radar target coordinates and the center coordinates of the visual pedestrian bounding box. Set matching distance threshold When the matching conditions are met, the millimeter-wave radar target and the pedestrian detected by vision are determined to be the same target; the matching conditions are: Furthermore, the millimeter-wave radar target coordinates fall within the coordinate range of the visual pedestrian bounding box; S103. For pedestrians identified as the same target, the visual detection results are corrected using millimeter-wave radar measurement data; The detection result correction includes: position correction and velocity correction; the corrected position is obtained by weighted summation of pedestrian position coordinates from visual detection and millimeter-wave radar; the corrected velocity is obtained by weighted summation of pedestrian velocity from visual detection and millimeter-wave radar.

[0007] Furthermore, in step S2, when a pedestrian is making violent movements, a deep learning-based trajectory prediction network is used to predict the pedestrian's trajectory. The process for determining violent movement is as follows: Set a threshold for motion acceleration and a threshold for the rate of change of turning angle. When the pedestrian's motion acceleration is greater than the threshold for motion acceleration or the rate of change of turning angle is greater than the threshold for the rate of change of angle, the pedestrian is considered to be in a state of violent movement.

[0008] Furthermore, the detailed process of step S3 is as follows: S301. Build a dynamic model and calculate the future state; model parameters include: state vector and control vector; Define sprinkler truck in The state vector at any given time includes: the horizontal and vertical coordinates of the sprinkler truck in the geodetic coordinate system, the heading angle of the sprinkler truck, the longitudinal velocity of the sprinkler truck, the longitudinal acceleration of the sprinkler truck, and the relative distance between the sprinkler truck and the nearest pedestrian; Define the control vector of the sprinkler truck actuator. The control vector includes: the front wheel steering angle and the change in acceleration. Construct a dynamic model that takes the current control vector and state vector as input and the next control vector and state vector as output; Set the prediction time domain of MPC as Step; calculate the future based on the dynamic model. The first time predicted Current state; S302. Construct a multi-objective weighting function; Multi-objective weighting function is derived from The objective function is obtained by summing up the sub-objective functions. Each sub-objective function consists of a functional term and its corresponding coefficient. The functional terms include: safety term, work efficiency term, steering comfort term, and acceleration comfort term. The safety factor is related to the predicted relative distance between the water truck and the nearest pedestrian; The operational efficiency is related to the predicted value of the longitudinal speed of the waterwheel; Steering comfort is related to the rate of change of the front wheel steering angle; Acceleration comfort is related to the longitudinal acceleration of the sprinkler truck; S303. Set constraints; constraints include: control quantity constraints, road boundary constraints, and pedestrian avoidance constraints. S304. Determine the optimal avoidance trajectory based on the optimization objective; The optimization objective is to minimize the objective function, and the optimal control sequence is obtained based on the optimization objective. The first control vector of the optimal control sequence is executed, which is the optimal avoidance trajectory for the current cycle; S305. After entering the next cycle, repeat steps S1 to S3 to obtain the optimal avoidance trajectory for the next cycle.

[0009] Furthermore, the multi-objective weighting function is adjusted according to the scenario; The scenarios are divided as follows: when the road width is less than the narrow width threshold, it is a narrow scenario; when the road width is greater than the wide width threshold, it is a wide scenario; when the distance between pedestrians and vehicles is less than the emergency distance threshold, it is an emergency scenario.

[0010] Furthermore, the water pressure adaptive model is related to the speed of the sprinkler truck, the relative distance between pedestrians and the sprinkler truck, and the road width.

[0011] Furthermore, the detailed process of step S5 is as follows: S501. Construct a relative motion coordinate system for vehicles and pedestrians; S502. Adjust the control quantity according to the PID algorithm to perform real-time adaptive adjustment of water pressure.

[0012] Furthermore, the operating modes of the water sprinkler truck include: normal watering mode and avoidance mode; When no pedestrian is detected, or the relative distance between the pedestrian and the sprinkler truck is greater than the safety threshold, the system is in normal sprinkler mode. At this time, the water pressure value is determined according to the current sprinkler truck speed, road width or work requirements, and the sprinkler task is executed. When a pedestrian is detected and the minimum relative distance between the pedestrian and the sprinkler truck predicted by MPC is less than the safety threshold, the avoidance mode is triggered. At this time, the optimal avoidance trajectory is generated according to the MPC algorithm, and the sprinkler truck is controlled to perform the avoidance action. At the same time, based on the real-time distance between the pedestrian and the sprinkler truck, the low target water pressure value is calculated through the water pressure model to reduce the water spraying range and intensity. After the avoidance is completed, the system will switch back to normal watering mode.

[0013] Furthermore, the control constraints are: the absolute value of the front wheel steering angle is less than the threshold value of the front wheel steering angle, and the longitudinal acceleration of the sprinkler truck is greater than the lower limit of the acceleration change and less than the upper limit of the acceleration change; Set road boundary constraint distances. The road boundary constraint is: the distance between the vehicle and the left and right boundaries is greater than or equal to the constraint distance. Set a safe distance threshold, and the pedestrian avoidance constraint is: the predicted relative distance between the sprinkler truck and the nearest pedestrian is greater than the safe distance threshold.

[0014] The present invention provides a method for dynamic water pressure control of intelligent sprinkler trucks based on MPC prediction, which has the following advantages over existing technologies: (1) Overcoming the problem of insufficient accuracy of a single sensor in complex environments in existing technologies, we adopt multi-sensor perception combined with deep learning and combined tracking algorithms to achieve accurate acquisition of pedestrian dynamic information and enhance the detection reliability in complex scenarios; (2) The trajectory planning prediction mechanism of model predictive control is adopted, and the avoidance strategy is automatically generated in combination with vehicle dynamic parameters, which upgrades the passive response to the active prevention and reduces the safety hazards caused by operational errors. (3) A water pressure model that integrates multiple factors such as driving speed, pedestrian distance, and road width is used to achieve real-time adjustment through closed-loop control; (4) Adjusting the optimization coefficients and multi-objective weighting functions according to the scenario can improve the scenario applicability of the method. Attached Figure Description

[0015] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of the embodiments of this application.

[0017] In the embodiments of this application, the terms "upper," "lower," "inner," "middle," "outer," "front," and "rear," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are mainly for better describing the embodiments of this application and their implementations, and are not intended to limit the indicated device, element, or component to having a specific orientation, or to be constructed and operated in a specific orientation. Furthermore, some of the above terms may be used to indicate other meanings besides orientation or positional relationship; for example, the term "upper" may also be used in some cases to indicate a certain dependency or connection relationship. For those skilled in the art, the specific meaning of these terms in the embodiments of this application can be understood according to the specific circumstances.

[0018] Furthermore, the terms "set up," "connect," and "fix" should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral structure; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or it can be an internal connection between two devices, components, or parts. Those skilled in the art can understand the specific meaning of the above terms in the embodiments of this disclosure according to the specific circumstances.

[0019] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the present application described herein can be implemented, for example, in orders other than those illustrated or described herein.

[0020] In this application, the terms "exemplarily" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplarily" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplarily" or "for example" is intended to present the relevant concepts in a specific manner.

[0021] Unless otherwise stated, the term "multiple" means two or more.

[0022] like Figure 1 As shown, a method for dynamic water pressure control of an intelligent sprinkler truck based on MPC prediction includes the following steps: S1. Perform data acquisition and preprocessing operations; The data includes: images of the vehicle's surroundings, distance, speed, and angle information of the target object, and the sprinkler truck's own position, heading angle, speed, and acceleration; The sprinkler truck is equipped with a 360° blind-spot-free multi-camera system and millimeter-wave radar. The multi-camera system includes a forward-looking long-range camera, a forward-looking near-range camera, left and right fisheye cameras, and a rear binocular camera. The multi-camera system provides image information about the vehicle's surroundings. The sprinkler truck is also equipped with an onboard T-BOX, an inertial navigation system (IMU), GPS, and millimeter-wave radar. The onboard T-BOX collects basic status data such as the sprinkler truck's positioning, vehicle networking, vehicle power supply, and equipment operating status. The IMU collects the sprinkler truck's heading angle, longitudinal acceleration, and longitudinal speed. The GPS collects the sprinkler truck's three-dimensional position coordinates in the geodetic coordinate system. At the same time, the vehicle's own speed sensor can also collect the sprinkler truck's speed and acceleration.

[0023] The images surrounding the vehicle undergo preprocessing operations, including: grayscale conversion, which converts color images to grayscale to reduce subsequent computation; Gaussian filtering to remove noise interference and improve image clarity; and Canny edge detection, which extracts edge features from the image to provide a feature basis for subsequent pedestrian detection.

[0024] The visual inspection results are corrected using millimeter-wave radar data. The correction process is as follows: S101. Transform the coordinate system of the millimeter-wave radar and extract pedestrian bounding box information; convert the target position output by the millimeter-wave radar from polar coordinates to Cartesian coordinates; extract the center coordinates of the pedestrian bounding box output by visual detection. The coordinates of the top left vertex of the bounding box's circumscribed rectangle are: The coordinates of the bottom right vertex are ; S102. Perform target matching and determination. The specific process is as follows: Calculate the Euclidean distance between the millimeter-wave radar target coordinates and the center coordinates of the visual pedestrian bounding box. The expression is as follows: (1); in, , These are the x and y coordinates of the millimeter-wave radar target, respectively. Set matching distance threshold When the matching conditions are met, the millimeter-wave radar target and the pedestrian detected by vision are determined to be the same target; the matching conditions are: Furthermore, the millimeter-wave radar target coordinates fall within the coordinate range of the visual pedestrian bounding box; S103. For pedestrians identified as the same target, the visual detection results are corrected using high-precision measurement data from millimeter-wave radar. The correction of the detection results includes: position correction and velocity correction; Let the horizontal and vertical coordinates be denoted as... The corrected positional expression is as follows: (2); in, These are the corrected pedestrian location coordinates. , These are the pedestrian location coordinates obtained from visual detection and millimeter-wave radar, respectively. , These are visual weights and radar weights, respectively; The corrected velocity expression is as follows: (3); in, For the corrected pedestrian speed, , The pedestrian speeds are determined by visual detection and millimeter-wave radar, respectively. Typically, the visual weight is 0.5, and the radar weight is 0.5. In extreme weather conditions such as rain, snow, or fog, the visual weight is adjusted to 0.1 and the radar weight to 0.9.

[0025] S2. Perform pedestrian detection and tracking operations; The pedestrian detection method used is the YOLOv5 algorithm; The target location detected by millimeter-wave radar is matched with the pedestrian bounding box detected by vision through a data association algorithm. If the radar target is within the error range of the pedestrian bounding box, it is determined to be the same target. The high-precision distance and velocity information provided by millimeter-wave radar is used to correct the pedestrian position and velocity data obtained by vision detection, thereby improving the detection accuracy in complex environments. The pedestrian tracking method uses a target tracking algorithm that combines Kalman filtering and Hungarian algorithm to continuously track detected pedestrians: Kalman filtering is used to predict the motion state of pedestrians, and Hungarian algorithm is used to perform data association on pedestrian targets in consecutive image frames to ensure that the same pedestrian is being tracked, and finally obtain the complete motion trajectory and real-time speed of the pedestrian. When pedestrians make rapid turns or sudden accelerations, traditional methods combining Kalman filtering and Hungarian algorithms may result in tracking loss. Deep learning-based trajectory prediction networks can more accurately predict pedestrian trajectories. By using the pedestrian state predicted by Kalman filtering as input to an LSTM network and combining it with historical trajectory data, the LSTM network learns the complex patterns and rules of pedestrian movement and outputs more accurate future trajectory predictions. In the data association stage, the prediction results of the LSTM network are matched with the detection target to improve tracking accuracy in cases of pedestrians making rapid movements. For example, when a pedestrian suddenly crosses the road, the improved tracking algorithm can track the pedestrian's movement in a timely and accurate manner, avoiding avoidance decision errors caused by tracking loss. The process for determining violent movement is as follows: Set a threshold for motion acceleration and a threshold for the rate of change of turning angle. When the pedestrian's motion acceleration is greater than the threshold for motion acceleration or the rate of change of turning angle is greater than the threshold for the rate of change of angle, the pedestrian is considered to be in a state of violent movement.

[0026] S3. Generate the optimal avoidance trajectory based on MPC. The detailed process is as follows: S301. Build a dynamic model and calculate the future state; model parameters include: state vector and control vector; Define the state vector of the sprinkler truck at time k. for: (4); in, , These represent the x and y coordinates of the water truck in the geodetic coordinate system. This refers to the heading angle of the water truck, which is the angle between the direction the water truck is facing and the due east direction. The longitudinal speed of the water truck For the longitudinal acceleration of the sprinkler truck, The relative distance between the water truck and the nearest pedestrian. It is the transpose symbol; Define the control vector of the sprinkler truck actuator for: (5); in, This is the front wheel steering angle, used to control the steering direction and amplitude. This is the change in acceleration; when it is greater than 0, it indicates acceleration, and when it is less than 0, it is equal to deceleration. The dynamic model reflects the variation of the state vector with the control vector, and its expression is as follows: (6); in, To control the cycle, The wheelbase of the water sprinkler truck is an inherent parameter of the vehicle itself. , The first The horizontal and vertical coordinates of the predicted position of the pedestrian at any given time; Set the prediction time domain of MPC as The control sequence can be obtained by taking 5 to 10 steps with a control period of 0.1 seconds. The expression is as follows: (7); The future state is calculated based on the dynamic model, and the expression is as follows: (8); in, For the first The first time predicted The state at time t can be obtained iteratively from the dynamic model, i.e., from formula (4); S302. Construct a multi-objective weighting function; The expression for the multi-objective weighting function is as follows: (9); in, , , , These are safety, operational efficiency, steering comfort, and acceleration comfort. , , , These are optimization coefficients for safety, work efficiency, steering comfort, and acceleration comfort. The optimization coefficients for safety and work efficiency are specifically adjusted based on the scenario, as follows: Set narrow width threshold A width of 6m indicates a narrow scenario. In this case, the safety optimization coefficient needs to be increased while the work efficiency optimization coefficient needs to be decreased to prioritize collision-free operation and allow for greater speed loss. Setting the narrowness adjustment coefficient to 1 / 5, the adjustment process for the safety optimization coefficient and the work efficiency optimization coefficient is as follows: (10); in, , These are the baseline values ​​for the safety coefficient and the work efficiency coefficient, respectively. , These are the adjustment values ​​for the safety coefficient and the work efficiency coefficient, respectively. Road width; Set width threshold The setting is 15m, meaning a road width greater than 15m is considered a wide scenario. The optimization coefficient for safety is reduced, while the optimization coefficient for work efficiency is increased to prioritize maintaining work speed and avoid obstacles through small turns. The wide scenario adjustment coefficient is set to 1 / 5. The adjustment process for the safety optimization coefficient and the reduction in the work efficiency optimization coefficient is as follows: (11); Set emergency distance threshold The distance is set to 3m, meaning an emergency scenario occurs when the distance between the pedestrian and the vehicle is less than 3m. In this case, only pedestrian avoidance is considered, so the expression of the target weighting function is modified as follows: (12); The expression for the security item is: (13); in, This is the safe distance threshold; The expression for the work efficiency term is: (14); in, For scene reference speed; The expression for the steering comfort term is: (15); in, For the first The rate of change of the front wheel steering angle at any given moment; The expression for accelerating comfort is: (16); S303. Set constraints; constraints include: control quantity constraints, road boundary constraints, and pedestrian avoidance constraints; control quantity constraints are: the first... The absolute value of the front wheel steering angle at time t is less than the threshold value of the front wheel steering angle, and the t is less than the threshold value of the front wheel steering angle. At any given moment, the longitudinal acceleration of the sprinkler truck is greater than the lower limit of the change in acceleration but less than the upper limit of the change in acceleration. Based on the prediction at time i, the relative distance between the sprinkler truck and the nearest pedestrian at time j should be greater than the safe distance threshold. The expression for the control constraint is as follows: (17); in, This is the threshold value for the front wheel steering angle, typically taken as 0.5 rad. , These are the lower and upper limits of the change in acceleration, typically taken as -3 m / s² and 1 m / s², respectively. The road boundary constraint distance is set to 0.5. The road boundary constraint is: the distance between the vehicle and the left and right boundaries is greater than or equal to 0.5m, that is, the distance between the vehicle and the left and right sides of the road should not be less than 0.5m. If the safe distance threshold is 0.5m, then the predicted relative distance between the sprinkler truck and the nearest pedestrian should be no less than 0.5m. S304. Determine the optimal avoidance trajectory based on the optimization objective; The optimization objective is to minimize the objective function. The optimal control sequence is then derived based on this objective. A fast quadratic programming algorithm is used to linearize nonlinear constraints in order to improve computation speed. The first control vector of the optimal control sequence is executed, which is the optimal avoidance trajectory for the current cycle; S305. After entering the next cycle, repeat steps S1 to S3 to obtain the optimal avoidance trajectory for the next cycle.

[0027] S4. Construct a water pressure adaptive model based on multiple working conditions to regulate the water pressure value; the mathematical expression of the water pressure adaptive model is: (18); in, For the target water pressure value, For the speed of the water truck, The distance between the pedestrian and the water truck is considered infinite when there are no pedestrians. , , These are the speed proportionality coefficient, distance proportionality coefficient, and road surface proportionality coefficient, respectively. The offset is a constant. Through multiple sets of standard operating condition tests, the optimal water pressure values ​​were collected under different driving speeds, road widths, and pedestrian distances. The results were obtained by fitting the values ​​using the least squares method. , , , The specific value.

[0028] S5. Use the PID algorithm for closed-loop control of water pressure. The detailed process is as follows: S501. Construct a relative motion coordinate system for the vehicle and pedestrians; the relative motion coordinate system is a planar coordinate system, with its origin at the geometric center of the sprinkler truck and its direction of travel as the coordinate system. The positive direction of the axis, perpendicular to the direction of travel, points to the right of the sprinkler truck. Positive direction of the axis; S502. Adjust the control quantity according to the PID algorithm to perform real-time adaptive adjustment of water pressure; The expression for the control variable is as follows: (19); in, For control quantities, the corresponding signals are pump speed control signals or valve opening control signals. , , These are the proportional, integral, and differential coefficients, respectively. This represents the deviation between the target water pressure value and the actual value.

[0029] S6. Switch the working mode of the sprinkler truck according to the actual road conditions; The working modes of the water sprinkler truck include: normal watering mode and avoidance mode; When no pedestrian is detected, or the relative distance between the pedestrian and the sprinkler truck is greater than the safety threshold, the system is in normal sprinkler mode. At this time, the water pressure value is determined according to the current sprinkler truck speed, road width, greening or cleaning needs, and the sprinkler task is executed. When a pedestrian is detected and the minimum relative distance between the pedestrian and the water truck is predicted to be less than a safety threshold using a relative motion model, an avoidance mode is triggered. At this time, the optimal avoidance trajectory is generated according to the MPC algorithm, and the water truck is controlled to perform avoidance actions such as deceleration and steering. At the same time, based on the real-time distance between the pedestrian and the water truck, a low target water pressure value is calculated using a water pressure model to reduce the water spraying range and intensity, and avoid splashing the pedestrian. Once the avoidance is completed, i.e., the relative distance between the pedestrian and the water truck is greater than the safety threshold, the system switches back to normal water spraying mode from the avoidance mode.

[0030] The above embodiments are merely preferred technical solutions of the present invention. Those skilled in the art, upon considering the specification and practicing the invention disclosed herein, will readily conceive of other embodiments of the present application. The embodiments of this application are intended to cover any variations, uses, or adaptations of the embodiments of this application that follow the general principles of the embodiments of this application and include common knowledge or customary technical means in the art not disclosed in the embodiments of this application. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of the embodiments of this application are indicated by the claims of this application.

Claims

1. A method for dynamic water pressure control of an intelligent sprinkler truck based on MPC prediction, characterized in that, Includes the following steps: S1. Perform data acquisition and preprocessing operations; S2. Perform pedestrian detection and tracking operations; S3. Generate the optimal avoidance trajectory based on MPC; S4. Construct a water pressure adaptive model based on multiple working conditions to regulate the water pressure value; S5. Use the PID algorithm to perform closed-loop control of water pressure; S6. Switch the working mode of the sprinkler truck according to the actual road conditions.

2. The method for dynamic water pressure control of an intelligent sprinkler truck based on MPC prediction according to claim 1, characterized in that, The data includes: Images of the vehicle's surroundings, including distance, speed, and angle information of the target object, as well as the sprinkler truck's own position, heading angle, speed, and acceleration.

3. The method for dynamic water pressure control of an intelligent sprinkler truck based on MPC prediction according to claim 2, characterized in that, In step S1, millimeter-wave radar data is used to correct the visual detection results. The correction process is as follows: S101. Transform the coordinate system of the millimeter-wave radar and extract pedestrian bounding box information; convert the target position output by the millimeter-wave radar from polar coordinates to Cartesian coordinates; S102. Perform target matching and determination. The specific process is as follows: Calculate the Euclidean distance between the millimeter-wave radar target coordinates and the center coordinates of the visual pedestrian bounding box. Set matching distance threshold When the matching conditions are met, the millimeter-wave radar target and the pedestrian detected by vision are determined to be the same target; the matching conditions are: Furthermore, the millimeter-wave radar target coordinates fall within the coordinate range of the visual pedestrian bounding box; S103. For pedestrians identified as the same target, the visual detection results are corrected using millimeter-wave radar measurement data; The detection result correction includes: position correction and velocity correction; the corrected position is obtained by weighted summation of pedestrian position coordinates from visual detection and millimeter-wave radar; the corrected velocity is obtained by weighted summation of pedestrian velocity from visual detection and millimeter-wave radar.

4. The method for dynamic water pressure control of an intelligent sprinkler truck based on MPC prediction according to claim 1, characterized in that, In step S2, when a pedestrian is making violent movements, a deep learning-based trajectory prediction network is used to predict the pedestrian's trajectory. The process for determining violent movement is as follows: Set a threshold for motion acceleration and a threshold for the rate of change of turning angle. When the pedestrian's motion acceleration is greater than the threshold for motion acceleration or the rate of change of turning angle is greater than the threshold for the rate of change of angle, the pedestrian is considered to be in a state of violent movement.

5. The method for dynamic water pressure control of an intelligent sprinkler truck based on MPC prediction according to claim 3, characterized in that, The detailed process of step S3 is as follows: S301. Build a dynamic model and calculate the future state; model parameters include: state vector and control vector; Define sprinkler truck in The state vector at any given time includes: the horizontal and vertical coordinates of the sprinkler truck in the geodetic coordinate system, the heading angle of the sprinkler truck, the longitudinal velocity of the sprinkler truck, the longitudinal acceleration of the sprinkler truck, and the relative distance between the sprinkler truck and the nearest pedestrian; Define the control vector of the sprinkler truck actuator. The control vector includes: the front wheel steering angle and the change in acceleration. Construct a dynamic model that takes the current control vector and state vector as input and the next control vector and state vector as output; Set the prediction time domain of MPC as Step; calculate the future based on the dynamic model. The first time predicted Current state; S302. Construct a multi-objective weighting function; Multi-objective weighting function is derived from The objective function is obtained by summing up the sub-objective functions. Each sub-objective function consists of a functional term and its corresponding coefficient. The functional terms include: safety term, work efficiency term, steering comfort term, and acceleration comfort term. The safety factor is related to the predicted relative distance between the water truck and the nearest pedestrian; The operational efficiency is related to the predicted value of the longitudinal speed of the waterwheel; Steering comfort is related to the rate of change of the front wheel steering angle; Acceleration comfort is related to the longitudinal acceleration of the sprinkler truck; S303. Set constraints; constraints include: control quantity constraints, road boundary constraints, and pedestrian avoidance constraints. S304. Determine the optimal avoidance trajectory based on the optimization objective; The optimization objective is to minimize the objective function, and the optimal control sequence is obtained based on the optimization objective. The first control vector of the optimal control sequence is executed, which is the optimal avoidance trajectory for the current cycle; S305. After entering the next cycle, repeat steps S1 to S3 to obtain the optimal avoidance trajectory for the next cycle.

6. The method for dynamic water pressure control of an intelligent sprinkler truck based on MPC prediction according to claim 5, characterized in that, Adjust the multi-objective weighting function according to the scenario; The scenarios are divided as follows: when the road width is less than the narrow width threshold, it is a narrow scenario; when the road width is greater than the wide width threshold, it is a wide scenario; when the distance between pedestrians and vehicles is less than the emergency distance threshold, it is an emergency scenario.

7. The method for dynamic water pressure control of an intelligent sprinkler truck based on MPC prediction according to claim 6, characterized in that, The water pressure adaptive model is related to the speed of the sprinkler truck, the relative distance between pedestrians and the sprinkler truck, and the road width.

8. The method for dynamic water pressure control of an intelligent sprinkler truck based on MPC prediction according to claim 7, characterized in that, The detailed process of step S5 is as follows: S501. Construct a relative motion coordinate system for vehicles and pedestrians; S502. Adjust the control quantity according to the PID algorithm to perform real-time adaptive adjustment of water pressure.

9. The method for dynamic water pressure control of an intelligent sprinkler truck based on MPC prediction according to claim 8, characterized in that, The working modes of the water sprinkler truck include: normal watering mode and avoidance mode; When no pedestrian is detected, or the relative distance between the pedestrian and the sprinkler truck is greater than the safety threshold, the system is in normal sprinkler mode. At this time, the water pressure value is determined according to the current sprinkler truck speed, road width or work requirements, and the sprinkler task is executed. When a pedestrian is detected and the minimum relative distance between the pedestrian and the sprinkler truck predicted by MPC is less than the safety threshold, the avoidance mode is triggered. At this time, the optimal avoidance trajectory is generated according to the MPC algorithm, and the sprinkler truck is controlled to perform the avoidance action. At the same time, based on the real-time distance between the pedestrian and the sprinkler truck, the low target water pressure value is calculated through the water pressure model to reduce the water spraying range and intensity. After the avoidance is completed, the system will switch back to normal watering mode.

10. The method for dynamic water pressure control of an intelligent sprinkler truck based on MPC prediction according to claim 5, characterized in that, The control constraints are: the absolute value of the front wheel steering angle is less than the threshold value of the front wheel steering angle, and the longitudinal acceleration of the sprinkler truck is greater than the lower limit of the acceleration change and less than the upper limit of the acceleration change; Set road boundary constraint distances. The road boundary constraint is: the distance between the vehicle and the left and right boundaries is greater than or equal to the constraint distance. Set a safe distance threshold, and the pedestrian avoidance constraint is: the predicted relative distance between the sprinkler truck and the nearest pedestrian is greater than the safe distance threshold.

Images