Automobile cleaning mode switching method based on self-adaptive model and intelligent switching system

Abstract

The invention discloses an automobile cleaning mode switching method and an intelligent switching system based on a self-adaptive model, and belongs to the technical field of automobile cleaning, and the system takes user operation logic as a core, and performs logic judgment and intelligent decision making on multi-source sensing signals in a programmed manner. By collecting multi-source data such as trigger opening and closing timestamps, pipeline pressure and flow and the like in real time and combining sliding window statistics and a K-means + HMM cascade algorithm, intelligent judgment of three hidden states of low-pressure wax spraying, interruption standby and double-click instructions is achieved; after a double-click switching instruction is judged, integrated switching of the water wax valve, the electromagnetic valve and the water pump can be completed within millisecond-level response. By embedding logic judgment and intelligent decision into a system control program, the system can realize efficient, personalized and reliable integrated cleaning mode switching without an external physical switch, so that the operation convenience and the user experience are greatly improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of automobile washing, and in particular relates to an automobile washing mode switching method and an intelligent switching system based on an adaptive model. Background Art

[0002] Currently, domestic car washing systems are primarily based on general industrial cleaning equipment. Precision cleaning equipment developed relatively late and suffers from insufficient technology, resulting in the industry as a whole still relying on simple models, with an urgent need for intelligent upgrades. Existing car washing methods can be categorized into manual car washing, automated brush washing, high-pressure water gun cleaning, steam cleaning, and waterless car washing. Manual car washing relies on manual operation, is inefficient, and consumes large amounts of water. Automatic brush washing can easily damage the paint and leaves many blind spots untouched. While high-pressure water gun cleaning is more efficient, switching modes requires an external switch or remote control, forcing the user to access the main unit for operation, reducing ease of use. While steam and waterless car washing can save water and improve environmental protection, they lack the ability to remove stubborn stains and are complex processes, hindering widespread adoption.

[0003] To enhance user experience, some high-pressure cleaners have a mode switch on the trigger or gun body of the cleaning gun. By detecting water pressure fluctuations or motor current fluctuations, the cleaning mode is switched without the user having to return to the main unit to operate, thus improving convenience. However, this method relies only on a single physical signal and cannot distinguish between user intentions and occasional fluctuations. It also lacks personalization and can lead to false touches, switching delays, or operator fatigue. In the field of spraying and cleaning, there are already adaptive control methods for drone spraying stability, which optimize spraying stability through multi-load coupling models and self-anti-disturbance controllers; there are also adaptive spraying systems based on images and PID variant algorithms, which can automatically select spray gun models and supply parameters in robotic systems. However, these methods mainly focus on spraying quality and path optimization, and have not yet addressed real-time intelligent switching of cleaning modes, let alone combining user usage habits and multi-source sensor data to achieve integrated control of the entire process.

[0004] The main problems currently exist are: mode switching relies on simple physical switches: it is difficult to accurately distinguish user intentions, prone to accidental touches and switching delays; lack of user personalized judgment: failure to perform intelligent switching based on user habit models, unable to meet the usage preferences of different users; insufficient multi-source signal fusion: failure to fully utilize multi-dimensional data such as trigger triggering, pressure and flow to determine the timing of switching; lack of integrated overall solutions: existing technologies mostly focus on subsystem optimization, and lack integrated implementation covering data collection, modeling, judgment and execution.

[0005] In response to the above technical bottlenecks, the present invention proposes an adaptive model-based automobile cleaning mode switching method and intelligent switching system. By building a user habit model and integrating multi-source sensor data, real-time, intelligent and personalized cleaning mode switching is achieved to improve cleaning efficiency, operational convenience and user experience. Summary of the Invention

[0006] Based on the above technical background, the existing technologies have the problems of mode switching relying on physical switches, lack of user personalized judgment, insufficient multi-source signal fusion, and lack of integration. The present invention proposes a car washing mode switching method and an intelligent switching system based on an adaptive model. The method collects multi-source data such as trigger opening and closing timestamps and pipeline pressure and flow in real time, combines sliding window statistics with the user habit model of K-means-HMM cascade, and realizes intelligent judgment of the three hidden states of low-pressure wax spraying, interrupt waiting and double-click instructions; the system includes a data acquisition module, a user habit modeling module, a mode switching judgment module, an execution module and a safe exit module, which can automatically complete the integrated switching from low-pressure water wax to high-pressure water or steam mode, and safely exit when the spray gun is idle for a long time.

[0007] The present invention provides a method for switching car washing modes based on an adaptive model, comprising the following steps:

[0008] S1: Data acquisition: The pressure curve P(t) and flow curve Q(t) are collected by the pressure sensor and flow sensor on the pipeline at a sampling frequency of ≥200Hz. The controller performs differential operation on the real-time collected pressure curve P(t). When |dP(t) / dt|≥ΔP, this moment is recorded as the trigger time t of the i-th trigger. i , forming a trigger time sequence T={t1,t2,...,t n}, all sampled data are time-stamped by the controller’s millisecond clock and then synchronously imported into the data buffer, where ΔP is the preset pressure threshold;

[0009] S2: User habit modeling: The controller sequentially calculates the time interval Δt between adjacent trigger moments i =t i+1 -t i , in the recent N s The average value μ and the sample standard deviation σ are calculated on the interval samples; the batch of interval sequences are input into the sliding window statistical K-means-HMM cascade algorithm, wherein the K-means process clusters the interval samples according to μ and σ. The cluster division includes the state transition probability matrix π of the three hidden states: low-pressure wax spraying state, interrupt waiting state and double-click instruction state, and generates the user habit model M u =(μ,σ,π);

[0010] S3: Determine the mode switching event: obtain the trigger interval Δt in real time during the new cleaning process j And with model M u Matching is performed based on the benchmark. If the match meets the trigger condition |Δt-μ|≤kσandC=2, it is determined that a switching instruction event E occurs.sw ; Where k is the preset tolerance ratio, C is the number of consecutive triggers within the detection window, and event E sw Indicates switching from low-pressure water wax mode to high-pressure water or steam mode;

[0011] S4: Mode conversion execution: When event E is detected sw When the controller executes the following steps in sequence: close the water wax solenoid valve, open the clean water solenoid valve or steam solenoid valve, and send a variable frequency speed control instruction to a single water pump;

[0012] S5: Safe Exit: If the spray gun is idle for a continuous period exceeding a preset threshold and no legal double-click event is detected, the controller executes a safe exit.

[0013] An adaptive model-based intelligent switching system for automobile washing modes is used to implement the above-mentioned adaptive model-based intelligent switching method for automobile washing modes. The system includes the following modules:

[0014] A single water pump assembly provides drive fluid pressure for all cleaning modes;

[0015] A multi-channel fluid pipeline assembly, including a water wax pipeline, a clean water pipeline, and a steam pipeline connected to the single water pump assembly, each pipeline being provided with a quick connector for mode switching;

[0016] a spray gun assembly, including a trigger, for turning the spray gun on and off;

[0017] Pipeline detection components, including pressure sensors and flow sensors respectively arranged on each fluid pipeline;

[0018] The solenoid valve assembly includes a water wax solenoid valve, a clean water solenoid valve, and a steam solenoid valve, which are respectively connected to corresponding pipelines and are used to switch the output fluid according to the controller instructions;

[0019] Controller component, integrating:

[0020] A data acquisition module is used to receive and time-stamp all raw signals from the spray gun assembly and the pipeline detection assembly using a millisecond clock and then import them into a data buffer;

[0021] The user habit modeling module is used to generate a user habit model M based on the latest N trigger triggering intervals;

[0022] The mode switching judgment module is used to match the latest trigger interval Δt with the model M in real time and determine the mode switching event E;

[0023] The mode conversion execution module is used to sequentially close the wax solenoid valve and open the clean water solenoid valve or the steam solenoid valve after event E is triggered, and at the same time issue a variable frequency speed control instruction to a single water pump component;

[0024] The safety exit module is used to close all solenoid valves and stop the operation of the single water pump assembly when the continuous idle time of the spray gun assembly exceeds a preset threshold and no legal double-click event is detected, so as to achieve safe exit.

[0025] As a preferred technical solution of the present invention, the data collection in step S1 includes the following steps:

[0026] The original signal is collected by the spray gun trigger stroke, pressure sensor and flow sensor at a sampling frequency of ≥200Hz;

[0027] In the controller, the above signals are respectively subjected to 5Hz low-pass filtering to remove noise, and the trigger opening and closing events are de-jittered using a 20ms anti-jitter algorithm;

[0028] The MCU millisecond-level system clock is used to timestamp the sampled data of each sensor, and the multi-source data is time-aligned to generate an accurate trigger time sequence T = {t1, t2, ..., t n}, pressure curve P(t) and flow curve Q(t).

[0029] As a technical preferred solution of the present invention, the user habit model in step S2 adopts an adaptive model updating method to continuously update parameters during the car washing process.

[0030] As a preferred technical solution of the present invention, the event E in step S3 and step S4 sw Specifically, the temperature judgment module is used to switch the automatic high-pressure water or steam mode. When the temperature is lower than 5 degrees Celsius, event E sw Represents steam mode, when the temperature is higher than 5 degrees Celsius, event E sw Stands for high pressure water cleaning mode.

[0031] As a preferred technical solution of the present invention, in step S3, a variable frequency speed control command is issued to a single water pump. The variable frequency speed control command adopts a segmented S-shaped pressurization curve algorithm, and the pressurization satisfies the following conditions:

[0032] At any time t, the pipeline pressure P(t) satisfies:

[0033]

[0034] Where P(t) is the pipeline output pressure at time t, in kgf / cm 2 ; P1 is the initial pressure at the end of wax spray mode; P2 is the target pressure of high pressure mode required after switching, the value is not less than 60kgf / cm 2 ; β is the steepness adjustment coefficient of the pressure curve, which is used to control the slope of the pressurization curve; t0 is the inflection point time of the S-shaped curve, indicating the moment when the pressurization rate is fastest.

[0035] At any time t, the pressure change rate after the pipeline is pressurized satisfies:

[0036]

[0037] Among them, P max_rate The maximum allowable pressure change rate preset for the system, in kgf / cm 2 s -1 The specific value of the maximum allowable pressure change rate is the maximum value that will not cause water hammer impact.

[0038] As a preferred technical solution of the present invention, the updating parameters of the user habit model in step S2 includes the following steps:

[0039] S2-1: In continuous N s Calculate the arithmetic mean μ and sample standard deviation σ on the trigger trigger interval samples;

[0040] S2-2: Perform K-means clustering on the trigger interval sample sequence based on μ and σ to obtain K cluster centers;

[0041] S2-3: For each cluster, the trigger event sequence in the cluster is constructed as a hidden Markov model containing three hidden states: low-pressure wax spraying state, interrupt waiting state and double-click instruction state;

[0042] S2-4: Store the μ, σ, and cluster centers together as the user habit model M n ;

[0043] S2-5: After N consecutive vehicles have completed their operations, when the arithmetic mean μ of the new batch of trigger intervals is new When the following conditions are met between μ and the original μ:

[0044] ∣μ new -μ|>σ

[0045] Update μ and σ by exponentially decaying the memory factor λ according to the following formula:

[0046] μ←λμ+(1-λ)μ new

[0047] σ←λσ+(1-λ)σ new

[0048] Based on the updated μ and σ, complete the user habit model M n Adaptive update.

[0049] Compared with the related prior art, the beneficial effects of the present invention are:

[0050] The system uses logical control to judge user operations as its core, and collects multi-source sensor signals such as trigger opening and closing timestamps, pipeline pressure and flow in real time. It programmatically uses sliding window statistics and K-means+HMM cascade algorithm to make logical judgments. It can accurately distinguish between the three hidden states of "low-pressure wax spraying", "interrupt standby" and "double-click switching", and can achieve intelligent mode switching without an external physical switch, significantly improving the level of automation and operational convenience.

[0051] The user habit model is dynamically updated internally, and the mean μ and standard deviation σ of the last N trigger intervals are calculated in real time. The judgment tolerance and detection window size are adaptively adjusted accordingly, fully taking into account the trigger rhythm and preferences of different users, and effectively avoiding false touches or missed switches caused by fixed thresholds.

[0052] The system continuously monitors the idle time of the spray gun during the control process and has a built-in safety exit module. Once the preset threshold is exceeded, all solenoid valves are automatically closed and the water pump is stopped. This eliminates hardware loss and safety risks caused by long-term idling or misjudgment, and improves the service life and operational reliability of the equipment.

[0053] Driven by intelligent decision-making, after double-clicking the switch command to confirm, the controller issues commands to close the water and wax valve, open the fresh water / steam valve, and perform variable frequency speed regulation within milliseconds, enabling fast and seamless switching between multiple modes. This eliminates the need to return to the main unit or manually operate the system, allowing users to focus on spray gun operation, significantly improving overall cleaning efficiency and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a structural diagram of a method for switching car washing modes based on an adaptive model according to the present invention;

[0055] In the figure: 21, single water pump assembly; 22, multi-channel fluid pipeline assembly; 23, spray gun assembly; 24, pipeline detection assembly; 25, solenoid valve assembly; 26, controller assembly; 261, data acquisition module; 262, user habit modeling module; 263, mode switching judgment module; 264, mode conversion execution module; 265, safe exit module.

[0056] Figure 2 The present invention is a flow chart of a car washing mode switching device based on an adaptive model. DETAILED DESCRIPTION

[0057] The present invention is further described below with reference to the accompanying drawings and examples. However, the present invention can be implemented in many different ways and should not be construed as limited to the illustrated embodiments; rather, these embodiments provide those skilled in the art with implementation methods that meet applicable legal requirements.

[0058] Example: Figure 1As shown, this embodiment uses a 48V DC brushless single water pump assembly 21 as the power source, and supplies liquid (steam) to the spray gun assembly 23 through three parallel fluid pipelines 22 (spray wax, clean water, steam). The three pipelines are connected to the water pump outlet at the inlet end and connected to the spray gun at the outlet end. The pressure and flow sensors of the pipeline detection assembly 24 are both 0.5% FS precision industrial-grade devices and are connected to the controller assembly 26 via the CAN bus. The controller uses a 32-bit MCU, integrated 4MspsADC, CAN and Wi-Fi, and the internal clock accuracy is 1ms:

[0059] like Figure 2 As shown, the operation of the vehicle washing mode switching method based on the adaptive model includes the following steps:

[0060] S1: Data acquisition: The pressure curve P(t) and flow curve Q(t) are collected by the pressure sensor and flow sensor on the pipeline at a sampling frequency of ≥200Hz. The controller performs differential operation on the real-time collected pressure curve P(t). When |dP(t) / dt|≥ΔP, this moment is recorded as the trigger time t of the i-th trigger. i , forming a trigger time sequence T={t1,t2,...,t n}, all sampled data are time-stamped by the controller’s millisecond clock and then synchronously imported into the data buffer, where ΔP is the preset pressure threshold;

[0061] S2: User habit modeling: The controller sequentially calculates the time interval Δt between adjacent trigger moments i =t i+1 -t i , in the recent N s The average value μ and the sample standard deviation σ are calculated on the interval samples; the batch of interval sequences are input into the sliding window statistical K-means-HMM cascade algorithm, wherein the K-means process clusters the interval samples according to μ and σ. The cluster division includes the state transition probability matrix π of the three hidden states: low-pressure wax spraying state, interrupt waiting state and double-click instruction state, and generates the user habit model M u =(μ,σ,π);

[0062] S3: Determine the mode switching event: obtain the trigger interval Δt in real time during the new cleaning process j And with model M u Matching is performed based on the benchmark. If the match meets the trigger condition |Δt-μ|≤kσandC=2, it is determined that a switching instruction event E occurs. sw , where: k is the preset tolerance ratio; C is the number of consecutive triggers within the detection window; event E sw Indicates switching from low-pressure water wax mode to high-pressure water or steam mode;

[0063] S4: Mode conversion execution: When event E is detected sw When the controller executes the following steps in sequence: close the water wax solenoid valve, open the clean water solenoid valve or steam solenoid valve, and send a variable frequency speed control instruction to a single water pump;

[0064] S5: Safe Exit: If the spray gun is idle for a continuous period exceeding a preset threshold and no legal double-click event is detected, the controller executes a safe exit.

[0065] like Figure 1 As shown, the hardware connection relationship is as follows:

[0066] The output port of the single water pump assembly 21 is connected to three fluid pipelines 22 via a three-way joint.

[0067] The ends of the multi-channel fluid pipeline 22 are respectively connected in series with a water wax solenoid valve, a clean water solenoid valve, and a steam solenoid valve (solenoid valve assembly 25), and the valve body adopts a DN6 quick-plug structure.

[0068] The spray gun assembly 23 is connected to the pipeline and is used to switch the spray gun.

[0069] The pressure and flow sensors of the pipeline detection assembly 24 are fixed to the end of each pipeline near the spray gun through an aviation plug to reduce signal delay.

[0070] The five major software modules within the controller component 26 all run on FreeRTOS tasks with a 5ms task cycle. Their priorities, from top to bottom, are data acquisition, mode switching determination, user behavior modeling, mode conversion execution, and safe exit. The functions of each control module are as follows:

[0071] Data Acquisition Module 261: Reads pressure and flow signals at a 200Hz sampling frequency; applies 5Hz IIR low-pass filtering for denoising; performs 20ms debounce on the trigger signal; and uniformly stamps the sampled values ​​with millisecond timestamps before writing them to a circular buffer. User Habit Modeling Module 262: A fixed sample of the 50 most recent trigger intervals is taken; the mean μ and standard deviation σ are calculated; and samples are classified using K-means into three categories: "low-pressure wax spraying," "interrupt waiting," and "double-click command." A hidden Markov model is trained for each sample category and combined into an overall user habit model M. An adaptive update is triggered after the same operator completes three vehicle operations, ensuring the model dynamically adjusts to the individual's pace. Mode Switching Decision Module 263: Reads the latest trigger interval in real time; compares it against the probability threshold of model M; outputs event E when the "double-click command" sub-model hits a set number of times n. Upon receiving event E, Mode Switching Execution Module 264 immediately closes the wax solenoid valve; opens the water or steam solenoid valve based on the ambient temperature and issues a variable frequency boost command for the water pump; implements soft start using a segmented S-shaped slope table to avoid water hammer. Safe Exit Module 265: Polls the spray gun idle timer. If the spray gun remains idle for more than 90 seconds and no valid double-click is detected during this time, all solenoid valves are closed and the pump is stopped. After the pump stops, an MQTT message is sent to the host computer, indicating "Safe Exit, Gun Removal and Recycling."

[0072] When a user operates the system of the present invention, the specific working steps are as follows:

[0073] Power-on self-test: After the controller is powered on, it will self-test the MCU, memory, solenoid valve coil resistance and sensor zero point;

[0074] Prepare to spray wax: The system is in low-pressure wax spray mode by default, and the operator can spray wax by pulling the trigger;

[0075] Double-click trigger: After finishing wax spraying, the operator quickly double-clicks the trigger with an interval of about 150ms;

[0076] Intelligent judgment: the judgment module 263 matches the detected Δt with the model M and confirms that event E has occurred;

[0077] Fast switching: the execution module 264 completes the closing of the water wax valve, the opening of the clean water / steam valve, and the boosting of the water pump within 30ms;

[0078] Normal flushing: The operator completes high-pressure water or steam flushing;

[0079] Task end: If there is no further operation within 90 seconds, the process will be safely exited, the pump valve will be closed, and the system will return to standby mode.

[0080] When a new employee at the same car wash replaced a new employee, their trigger operation pace slowed significantly. The system automatically relaxed tolerances based on the degree of μ and σ drift and recalibrated the three cluster centers after three cars. This process eliminated the need for manual recalibration, maintaining high recognition accuracy.

[0081] The device in this embodiment was deployed in an outdoor car wash and tested continuously on 200 vehicles. The results showed that the average response time for switching from wax spraying to high-pressure mode was 28ms, the pattern recognition accuracy was 98.6%, and there were no cases of water hammer or pump idling, indicating that the hardware was intact.

[0082] In actual car washing operations, the device of this embodiment defaults to low-pressure wax spraying mode after power-on self-test. The operator simply picks up the spray gun and pulls the trigger to evenly apply wax. After waxing is complete, gently "double-pull" the trigger (with an interval of approximately 150ms between pulls). The controller, leveraging millisecond-level timestamps and an adaptive user habit model, instantly recognizes the switching command, simultaneously closing the water and wax valve, opening the water or steam valve, and driving the water pump to steadily increase pressure according to an S-shaped pressure increase curve. High-pressure water (or steam) is then ejected to complete the rinse / steam drying process. If the spray gun is lowered for 90 seconds without any operation, the system automatically shuts off the water pump and all solenoid valves for a safe exit. The entire process requires no trips back and forth to the main unit or physical levers, allowing the operator to remain focused on the spray gun action, achieving a seamless "pull-double-pull-rinse" process. Simultaneously, the software self-learns the individual's rhythm in real time, preventing accidental touches while ensuring extremely fast response. Combined with multiple sensors and valve and pump protection, this not only improves car washing speed and cleaning quality, but also significantly reduces energy consumption, equipment wear, and maintenance costs.

[0083] The above embodiments merely illustrate several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, and all such variations and improvements fall within the scope of protection of the present invention.

Claims

1. A method for switching car washing modes based on an adaptive model, characterized by: The method comprises the following steps: S1: Data acquisition: The pressure curve P(t) and flow curve Q(t) are collected by the pressure sensor and flow sensor on the pipeline at a sampling frequency of ≥200Hz. The controller performs differential operation on the real-time collected pressure curve P(t). When |dP(t) / dt|≥ΔP, this moment is recorded as the trigger time t of the i-th trigger. i , forming a trigger time sequence T={t1,t2,...,t n }, all sampled data are time-stamped by the controller’s millisecond clock and then synchronously imported into the data buffer, where ΔP is the preset pressure threshold; S2: User habit modeling: The controller sequentially calculates the time interval Δt between adjacent trigger moments i =t i+1 -t i , in the recent N s The average value μ and the sample standard deviation σ are calculated on the interval samples; the batch of interval sequences are input into the sliding window statistical K-means-HMM cascade algorithm, wherein the K-means process clusters the interval samples according to μ and σ. The cluster division includes the state transition probability matrix π of the three hidden states: low-pressure wax spraying state, interrupt waiting state and double-click instruction state, and generates the user habit model M u =(μ,σ,π); S3: Determine the mode switching event: obtain the trigger interval Δt in real time during the new cleaning process j And with model M u Matching is performed based on the benchmark. If the match meets the trigger condition |Δt-μ|≤kσandC=2, it is determined that a switching instruction event E occurs. sw ; Where k is the preset tolerance ratio, C is the number of consecutive triggers within the detection window, and event E sw Indicates switching from low-pressure water wax mode to high-pressure water or steam mode; S4: Mode conversion execution: When event E is detected sw When the controller executes the following steps in sequence: close the water wax solenoid valve, open the clean water solenoid valve or steam solenoid valve, and send a variable frequency speed control instruction to a single water pump; S5: Safe Exit: If the spray gun is idle for a continuous period exceeding a preset threshold and no legal double-click event is detected, the controller executes a safe exit.

2. The method for switching car washing modes based on an adaptive model according to claim 1, characterized in that: The data collection in step S1 includes the following steps: The original signal is collected by the spray gun trigger stroke, pressure sensor and flow sensor at a sampling frequency of ≥200Hz; In the controller, the above signals are respectively subjected to 5Hz low-pass filtering to remove noise, and the trigger opening and closing events are de-jittered using a 20ms anti-jitter algorithm; The MCU millisecond-level system clock is used to timestamp the sampled data of each sensor, and the multi-source data is time-aligned to generate an accurate trigger time sequence T = {t1, t2, ..., t n }, pressure curve P(t) and flow curve Q(t).

3. The method for switching car washing modes based on an adaptive model according to claim 1, characterized in that: In step S2, the user habit model uses an adaptive model updating method to continuously update parameters during the car washing process.

4. The method for switching car washing modes based on an adaptive model according to claim 1, characterized in that: Event E in step S3 and step S4 sw Specifically, the temperature judgment module is used to switch the automatic high-pressure water or steam mode; when the temperature is lower than 5 degrees Celsius, event E sw Represents steam mode; when the temperature is higher than 5 degrees Celsius, event E sw Stands for high pressure water cleaning mode.

5. The method for switching car washing modes based on an adaptive model according to claim 1, characterized in that: In step S3, a variable frequency speed control command is issued to a single water pump. The variable frequency speed control command adopts a segmented S-shaped pressurization curve algorithm, and the pressurization satisfies the following conditions: At any time t, the pipeline pressure P(t) satisfies: Where P(t) is the pipeline output pressure at time t, in kgf / cm 2 ; P1 is the initial pressure at the end of wax spray mode; P2 is the target pressure of high pressure mode required after switching, the value is not less than 60kgf / cm 2 β is the steepness adjustment coefficient of the pressure curve, which is used to control the slope of the pressure curve; t0 is the inflection point time of the S-shaped curve, indicating the moment when the pressure increase rate is the fastest; At any time t, the pressure change rate after the pipeline is pressurized satisfies: Among them, P max_rate The maximum allowable pressure change rate preset for the system, in kgf / cm 2 s -1 The specific value of the maximum allowable pressure change rate is the maximum value that will not cause water hammer impact.

6. The method for switching car washing modes based on an adaptive model according to claim 3, characterized in that: The updating of the parameters of the user habit model in step S2 includes the following steps: S2-1: In continuous N s Calculate the arithmetic mean μ and sample standard deviation σ on the trigger trigger interval samples; S2-2: Perform K-means clustering on the trigger interval sample sequence based on μ and σ to obtain K cluster centers; S2-3: For each cluster, the trigger event sequence in the cluster is constructed as a hidden Markov model containing three hidden states: low-pressure wax spraying state, interrupt waiting state and double-click instruction state; S2-4: Store the μ, σ, and cluster centers together as the user habit model M n ; S2-5: After N consecutive vehicles have completed their operations, when the arithmetic mean μ of the new batch of trigger intervals is new When the following conditions are met between μ and the original μ: ∣μ new -μ∣>σ Update μ and σ by exponentially decaying the memory factor λ according to the following formula: μ←λμ+(1-λ)μ new σ←λσ+(1-λ)σ new Based on the updated μ and σ, complete the user habit model M n Adaptive update.

7. An intelligent switching system for car washing modes based on an adaptive model, characterized by: The switching system is used to implement the switching method according to any one of claims 1 to 6, and the switching system includes: a single water pump assembly (21) for providing driving fluid pressure for all cleaning modes; A multi-channel fluid pipeline assembly (22) includes a water wax pipeline, a clean water pipeline, and a steam pipeline connected to the single water pump assembly (21), each pipeline being provided with a quick connector for mode switching; A spray gun assembly (23), including a trigger for turning the spray gun on and off; A pipeline detection assembly (24) includes a pressure sensor and a flow sensor respectively arranged on each fluid pipeline; A solenoid valve assembly (25), comprising a wax solenoid valve, a clean water solenoid valve and a steam solenoid valve, each connected to a corresponding pipeline and used to switch the output fluid according to the controller instruction; A controller assembly (26) integrating: A data acquisition module (261) is used to receive and time-stamp all raw signals from the spray gun assembly (23) and the pipeline detection assembly (24) using a millisecond clock and then import them into a data buffer; A user habit modeling module (262) for generating a user habit model M based on the most recent N trigger triggering intervals; A mode switching judgment module (263) is used to match the latest trigger interval Δt with the model M in real time to determine the mode switching event E; A mode conversion execution module (264) is used to sequentially close the wax solenoid valve and open the clean water solenoid valve or the steam solenoid valve after event E is triggered, and simultaneously issue a variable frequency speed control instruction to a single water pump component (21); The safety exit module (265) is used to close all solenoid valves and stop the operation of the single water pump assembly (21) when the continuous idle time of the spray gun assembly (23) exceeds a preset threshold and no legal double-click event is detected, so as to achieve safety exit.

Images