Control method and system for preventing boiler overpressure explosion in door-to-door steam car washing

Abstract

The invention discloses a control method and system for preventing boiler overpressure explosion in door-to-door steam car washing, the system comprises a boiler, a water pump, a pressure sensor, a gas solenoid valve and an electronic control unit, the pressure sensor collects the pressure of the boiler in real time, the Mel spectrum feature extraction and dynamic time warping (DTW) algorithm is used for monitoring the pressure change trend in real time, and the pressure change trend is monitored in real time. Abnormality identification and safety control of the stages of preheating, heating, standby, working and shutdown of the boiler are achieved. The problem that the sensor is frozen or breaks down in the low-temperature environment is effectively solved, the accuracy, safety and response speed of boiler explosion prevention are remarkably improved, and safe operation of equipment is ensured.

Description

Technical Field

[0001] The present invention belongs to the field of steam car washing, and in particular relates to a control method and system for preventing boiler overpressure explosion during door-to-door steam car washing. Background Art

[0002] In recent years, with growing environmental awareness and market demand for efficient and convenient services, door-to-door steam car washing has gained widespread attention. Traditional car washing methods often rely on large amounts of water and chemical cleaning agents. Door-to-door steam car washing, on the other hand, utilizes high-temperature, high-pressure steam to clean the vehicle's surface, effectively removing stains while significantly reducing water waste and environmental pollution. Furthermore, door-to-door steam car washing offers advantages such as easy installation, flexible operation, and a wide range of services. It is particularly suitable for high-density urban areas and locations with high environmental requirements, and its market prospects are promising. Currently, the core component of door-to-door steam car washing equipment is the boiler. Boilers generate high-temperature steam through gas or oil heating, and their operating status is directly related to the safety and reliability of the equipment. To ensure safe operation under high temperature and high pressure, existing technologies primarily use mechanical pressure and temperature sensors to monitor internal boiler parameters and implement overpressure and overtemperature protection measures. If the boiler pressure or temperature exceeds preset safety limits, the system automatically disconnects the gas supply or stops heating, thereby preventing overpressure and explosion accidents.

[0003] In the existing technology, some patents have proposed different solutions for boiler safety monitoring. For example, Patent No. CN201910923537.8 discloses a boiler liquid level detection method, device, storage medium, and control equipment. The method detects the pressure, temperature, and water quality inside the boiler, and calculates the liquid level in combination with the reference density of water and the boiler shape parameters. This achieves online detection of the liquid level and improves equipment safety and internal space utilization to a certain extent. Patent No. CN202410403851.4 discloses a safety monitoring platform for marine boilers. The method mainly monitors the operating status of marine boilers, collects real-time working status data and standard working status data inside the boiler, analyzes the boiler operation status through the working safety assessment coefficient, and then determines whether to generate an early warning information and, if necessary, generates a stop operation instruction to ensure the safe operation of the boiler. Patent No. CN202111226718.9 discloses a boiler safety evaluation method, device and system. Its technical solution analyzes the probability of various types of boiler explosion accidents, and combines real-time operating parameters and inspection information to provide early warning of boiler explosion risks, thereby achieving evaluation and control of the boiler's safety status.

[0004] While the above technologies have improved boiler monitoring capabilities and safety to a certain extent, they still have the following shortcomings: Inadequate environmental adaptability: Existing technologies primarily rely on traditional mechanical pressure and temperature sensors for monitoring. These sensors are prone to freezing or failure in harsh environments such as low temperatures and humidity, resulting in inaccurate data collection and a failure to timely reflect the true internal boiler status, posing potential safety hazards. Simple data processing methods: Existing systems typically use fixed threshold detection or simple sensor data comparison methods. They lack in-depth time-frequency analysis and pattern recognition for dynamically changing nonlinear signals such as boiler pressure, making them difficult to address abnormalities caused by operator negligence, equipment aging, or environmental changes. Limited safety response speed: Traditional safety control strategies are often based on single-data monitoring. When a sensor or pipeline malfunctions, the system cannot accurately determine the boiler's status, resulting in a slow response and inability to implement effective safety measures immediately, posing an explosion risk. Inadequate full-process safety closed-loop: Existing technologies often focus on monitoring the boiler's operating phase, but lack comprehensive monitoring of dynamic pressure changes during preheating, standby, and shutdown phases, failing to establish a comprehensive safety closed-loop from startup to shutdown.

[0005] Based on the above defects, there is an urgent need for a new type of boiler safety monitoring and control method and system, which should meet the following technical requirements: high-precision real-time monitoring: in various working environments, including low temperature and high humidity conditions, boiler pressure data can be collected in real time and accurately to ensure the integrity and accuracy of the data; advanced signal processing technology: the pressure signal is converted into a Mel spectrum feature matrix by time-frequency analysis, and the real-time data is nonlinearly matched with the standard benchmark through the dynamic time warping (DTW) matching algorithm to achieve fine identification of the status of the boiler at each stage; full-process safety protection: continuous monitoring is carried out in each stage of low-fire preheating, high-fire heating, standby, operation and shutdown. , adopts a double preheating judgment mechanism to ensure that all key components of the boiler are in normal condition before high-fire heating, and can immediately cut off the gas supply, activate the safety pressure relief device and issue an alarm signal when an abnormality is found; system reliability and stability: a monitoring system with redundant data acquisition and error detection functions is designed. In the event of sensor failure, communication interruption or data processing abnormality, it can automatically switch to the backup plan to ensure the continuous and stable operation of the system; compact, lightweight and miniaturized structure: under the premise of ensuring safety, the equipment structure is optimized as much as possible, the volume and weight of the boiler system are reduced to meet the requirements of door-to-door steam car washing equipment for lightness and portability, while reducing manufacturing and maintenance costs. Summary of the Invention

[0006] Based on the deficiencies of the existing technologies in the above technical background, the present invention proposes a control method and system for preventing boiler overpressure explosion during on-site steam car washing. The method monitors the boiler pressure signal in real time in stages through dual low-fire preheating discrimination, Mel spectrum feature extraction, and dynamic time warping matching algorithm, and uses a preset similarity threshold to accurately compare the real-time collected data with the reference template to achieve rapid identification and processing of abnormal conditions; the system consists of hardware such as the boiler body, pressure sensor, water pump, gas solenoid valve, electronic control unit, and operation panel, and integrates advanced signal processing and safety control algorithms in the control unit to achieve full-process safety closed-loop monitoring from preheating, high-fire heating, standby, operation to shutdown, thereby effectively reducing the risk of boiler overpressure explosion caused by sensor failure, data anomaly, or negligent operation, improving the overall safety and operational reliability of the equipment, and ensuring the stability and economic benefits of on-site steam car washing operations.

[0007] The present invention proposes a control method for preventing boiler overpressure explosion during on-site steam car washing, comprising the following steps:

[0008] S1: Initialize the device, start the control unit, load the Mel spectrum reference template and dynamic time warping matching algorithm parameters for each stage, and perform communication and hardware self-test on the pressure sensor, water pump, and gas solenoid valve;

[0009] S2: First low-fire preheating judgment: The control unit drives the gas solenoid valve to the low-fire state to preheat the boiler and collects pressure data in real time. The collected pressure data is preprocessed to obtain a pressure time series. Short-time Fourier transform and Mel scaling conversion are used to generate Mel spectrum features. This feature matrix is matched with the first preheating benchmark C1 through dynamic time warping for similarity. If the similarity result meets the threshold, the next step is entered; otherwise, heating is stopped and an alarm is issued.

[0010] S3: Second low-fire preheating determination: Repeat the acquisition and Mel spectrum feature extraction process in step S2. Use dynamic time warping to calculate the distance between the obtained second preheating Mel spectrum feature and the second preheating benchmark C2 and obtain the similarity. If the similarity result meets the threshold, subsequent high-fire heating is allowed. Otherwise, heating is stopped and an alarm is issued.

[0011] S4: High-fire heating and standby. If step S3 is qualified, the control unit switches the gas solenoid valve to the high-fire state to heat the boiler to the working pressure. By collecting pressure data in real time and generating Mel-spectrum features, the control unit performs dynamic time warping matching with the high-fire heating benchmark C3 to monitor the boiler pressure trend. If the matching result deviates from the threshold, the gas solenoid valve is immediately closed and an alarm is issued. When the boiler reaches the set pressure, it enters the standby state and maintains a stable pressure. During this period, the pressure data is continuously monitored and matched with the standby benchmark C4.

[0012] S5: Working mode: The user turns on the steam car wash mode and uses the steam gun. The control unit periodically collects pressure data to generate Mel-spectrum features, performs dynamic time warping matching with the working benchmark C5, and determines whether there is a pressure anomaly. If the matching result is lower than the set threshold, the gas solenoid valve is quickly closed and an alarm is issued. If the matching result is normal, combustion continues to be maintained to output stable steam.

[0013] S6: Shutdown monitoring. After the user finishes washing the car, the shutdown command is executed. The control unit stops combustion and collects boiler pressure data. The obtained pressure time series in the shutdown phase is converted into Mel spectrum and dynamically time-warped with the shutdown benchmark C6. If it is detected that the pressure cannot drop normally or the curve deviates from the threshold during the shutdown process, an alarm is issued and the system is locked. Otherwise, the shutdown is completed after the pressure drops to a safe range and the operating data is stored in the control unit or remote database.

[0014] A control system for preventing boiler overpressure explosion during on-site steam car washing is provided. The control system is used to implement the control method described above. The control system includes: a boiler body, a pressure sensor, a water pump, a gas solenoid valve, an operation panel, and an electronic control unit. The specific functions of each unit are as follows:

[0015] The boiler body is used to carry and heat the incoming water flow to generate high-temperature steam;

[0016] The pressure sensor is fixedly installed on the high-pressure pipeline of the boiler body and outputs a real-time pressure signal, including an analog-to-digital converter and a buffer module;

[0017] The water pump is connected to the boiler body to send external water into the boiler body for heating;

[0018] The gas solenoid valve is used to control the heating of the boiler by gas or oil, and switches between low fire and high fire according to the instructions issued by the electronic control unit;

[0019] The operation panel is used to input steam car wash mode selection and shutdown instructions and display the operating status. The electronic control unit has a built-in algorithm module and is respectively communicated with the pressure sensor, water pump, gas solenoid valve and operation panel;

[0020] The algorithm module performs short-time Fourier transform and Mel scale conversion on the collected boiler pressure data to generate a Mel spectrum feature matrix, and uses dynamic time warping to perform similarity matching with the benchmark templates of each stage.

[0021] As a technical preferred solution of the present invention: the pressure data acquisition process in steps S2-S5 includes the following steps: first, continuously acquiring pressure signals from the pressure sensor at a fixed sampling frequency through an analog-to-digital converter, then performing filtering, denoising and normalization processing on the acquired original signals and storing them in a cache module of the control unit according to time frames, and finally transmitting the pressure data in the cache to the algorithm module for Mel spectrum conversion and dynamic time warping matching.

[0022] As a preferred technical solution of the present invention, the method for generating the Mel spectrum features in steps S2-S5 includes the following steps:

[0023] First, the pressure time series after filtering and normalization is framed and windowed, and then short-time Fourier transform is performed to obtain the time-frequency amplitude information P(t,ω). The calculation formula of short-time Fourier transform is:

[0024]

[0025] Where p(τ) represents the value of the pressure signal at time τ, w(·) represents the window function, N is the number of sampling points in each frame, the window function w(t-τ) is used to reduce the frame edge effect, t represents the time frame index, P(t,ω) represents the complex amplitude at the t-th frame and angular frequency ω, and j is the imaginary unit;

[0026] Then, we take the modulus of P(t,ω) to get the amplitude spectrum, and then map the linear frequency f to the Mel frequency M(f). The calculation formula of this association is:

[0027]

[0028] Where M(f) represents the corresponding Mel scale frequency value;

[0029] Then, the energy of P(t,ω) is integrated on the Mel scale using a pre-set Mel filter bank to obtain the energy vector of each time frame.

[0030] Finally, logarithmic compression is performed on the energy vector to generate a two-dimensional Mel spectrum feature matrix. The calculation formula of the matrix is:

[0031] X∈R T×F

[0032] Where X represents the Mel spectrum feature matrix, T is the total number of time frames, and F is the number of Mel filter banks.

[0033] As a preferred technical solution of the present invention, the dynamic time warping matching process includes the following steps:

[0034] T1: Mel spectrum feature matrix X∈RT×F Decompose into a time frame vector x of real-time data acquisition t , T represents the total time frames of real-time data, and x t =[x t,1 ,x t,2 ,…,x t,F ] for each x t,f Represents the energy value of the f-th Mel filter output in the t-th frame, where f = 1, 2, ..., F, F represents the total number of Mel filter groups; the Mel spectrum feature matrix Y∈R of the reference state obtained in advance and preprocessed U×F Decomposed into the time frame vector y of the benchmark data u , u represents the time frame index of the benchmark data, U represents the collection of time frames, where the benchmark state includes [C1, C2, C3, C4, C5];

[0035] T2: Define the local distance function. The specific calculation formula is:

[0036]

[0037] Among them, w f represents the weighting coefficient of the f-th Mel filter, and log represents the natural logarithm function;

[0038] T3: Construct a cumulative distance matrix. The specific calculation formula is:

[0039] D(t,u)=δ(x t ,y u )+min{D(t-1,u),D(t,u-1),D(t-1,u-1)}

[0040] Where D(t,u) represents the minimum cumulative matching distance between the t-th frame of real-time data and the u-th frame of reference data. The initial conditions are set as D(0,0)=0, D(t,0)=+∞, D(0,u)=+∞, where +∞ represents a sufficiently large positive number.

[0041] T4: Calculate the normalized matching distance. The specific calculation formula is:

[0042]

[0043] Where D(T,U) represents the value of the cumulative distance matrix at the last frame of real-time data and the last frame of reference data;

[0044] T5: Convert the normalized matching distance into similarity. The specific similarity calculation formula is:

[0045] Sim(X,Y)=exp(-αDist DTW (X,Y)

[0046] Among them, α is a positive definite parameter used to adjust the mapping relationship between distance and similarity.

[0047] As a preferred technical solution of the present invention, the system adopts different processing methods for different exceptions, which are specifically divided into:

[0048] When the pressure sensor output data is lost, fluctuates abnormally, or the value exceeds the preset reasonable range, the control unit automatically identifies it as a sensor failure through real-time monitoring and redundant data collection, and immediately performs backup sensor calibration, data interpolation compensation, and continuous abnormality confirmation. It then closes the gas solenoid valve, activates the safety pressure relief device, and implements system shutdown operations;

[0049] When the water pump status detection result does not match the preset water supply status or detects a water supply interruption, the control unit immediately interrupts the heating operation, stops the gas supply, and starts the standby water pump detection program and alarm prompts;

[0050] When a calculation error, algorithm anomaly, or real-time matching similarity falls below a preset threshold during Mel spectrum feature conversion or dynamic time warping matching, the control unit automatically determines it as a data processing anomaly, immediately interrupts the current stage of operation, saves the abnormal data log, closes the gas solenoid valve, activates the safety pressure relief device, and issues an alarm signal;

[0051] When signal interruption, data transmission error or interference occurs in the communication interfaces between the pressure sensor, gas solenoid valve, operation panel and control unit, the system automatically starts the redundant communication channel, performs communication retry, and forces an emergency shutdown after continuous retry failures, closes the gas solenoid valve, starts the safety pressure relief device and records the fault log.

[0052] Compared with the related prior art, the present invention has the following beneficial effects:

[0053] Improved safety: This invention uses dual low-fire preheating discrimination and real-time monitoring based on Mel-spectrum feature extraction and dynamic time warping matching algorithms to perform precise comparisons of boiler pressure signals at each stage (preheating, high-fire, standby, operation, and shutdown). Once an anomaly is detected, the gas supply is quickly cut off and safety pressure relief measures are initiated, effectively preventing overpressure explosion accidents caused by frozen sensors, negligent operation, or equipment failure.

[0054] Strong real-time response capability: The system can continuously and in real time collect and process pressure data, quickly identify and respond to abnormal conditions, ensure the completion of safe closed-loop control in a short time, and improve the timeliness of accident warning and handling;

[0055] Excellent environmental adaptability: By adopting advanced signal processing technology to convert pressure signals into Mel spectrum features, accurate data collection and processing can be ensured even in harsh environments such as low temperature and humidity, thereby enhancing the stability and reliability of the system under various working conditions;

[0056] System structure optimization: The present invention optimizes the internal structure of the equipment through lightweight and miniaturized design, reduces system manufacturing and maintenance costs, and saves equipment space, meeting the portability and compactness requirements of door-to-door steam car washing equipment;

[0057] Improved overall economic benefits: Through full-process safety monitoring and accurate anomaly detection, economic losses caused by equipment downtime and accident repairs are reduced, while equipment operating efficiency and user experience are improved, thereby enhancing market competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 This is a flow chart of a control method for preventing boiler overpressure explosion in door-to-door steam car washing according to the present invention;

[0059] Figure 2 This is a structural diagram of a control system for preventing boiler overpressure explosion in a door-to-door steam car wash system according to the present invention;

[0060] Figure 3 It is a schematic diagram of a real-time pressure data curve of the boiler of the door-to-door steam car washing equipment under actual working conditions provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0061] 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.

[0062] Example 1: According to Figure 1 As shown, this embodiment provides a control method for preventing boiler overpressure explosion during door-to-door steam car washing, and the specific steps are as follows:

[0063] S1: Initialize the device, start the control unit and load the Mel spectrum reference template and dynamic time warping matching algorithm parameters for each stage, and perform communication and hardware self-test on the pressure sensor, water pump and gas solenoid valve; first, continuously obtain the pressure signal from the pressure sensor through the analog-to-digital converter at a fixed sampling frequency, then perform filtering, denoising and normalization processing on the obtained raw signal and store it in the cache module of the control unit according to the time frame, and finally transmit the pressure data in the cache to the algorithm module for Mel spectrum conversion and dynamic time warping matching.

[0064] S2: First low-fire preheating judgment: The control unit drives the gas solenoid valve to the low-fire state to preheat the boiler and collects pressure data in real time. The collected pressure data is preprocessed to obtain a pressure time series. Short-time Fourier transform and Mel scaling conversion are used to generate Mel spectrum features. This feature matrix is matched with the first preheating benchmark C1 through dynamic time warping for similarity. If the similarity result meets the threshold, the next step is entered; otherwise, heating is stopped and an alarm is issued.

[0065] S3: Second low-fire preheating determination: Repeat the acquisition and Mel spectrum feature extraction process in step S2. Use dynamic time warping to calculate the distance between the obtained second preheating Mel spectrum feature and the second preheating benchmark C2 and obtain similarity. If the similarity result meets the threshold, subsequent high-fire heating is allowed; otherwise, heating is stopped and an alarm is issued. The generation method of the Mel spectrum feature includes the following steps:

[0066] First, the pressure time series after filtering and normalization is framed and windowed, and short-time Fourier transform is performed to obtain the time-frequency amplitude information P(t,ω). The calculation formula of short-time Fourier transform is:

[0067]

[0068] Where p(τ) represents the value of the pressure signal at time τ, w(·) represents the window function, N is the number of sampling points in each frame, the window function w(t-τ) is used to reduce the frame edge effect, t represents the time frame index, P(t,ω) represents the complex amplitude at the t-th frame and angular frequency ω, and j is the imaginary unit;

[0069] Then, we take the modulus of P(t,ω) to get the amplitude spectrum, and then map the linear frequency f to the Mel frequency M(f). The calculation formula of this association is:

[0070]

[0071] Where M(f) represents the corresponding Mel scale frequency value;

[0072] Then, the energy of P(t,ω) is integrated on the Mel scale using a pre-set Mel filter bank to obtain the energy vector of each time frame.

[0073] Finally, logarithmic compression is performed on the energy vector to generate a two-dimensional Mel spectrum feature matrix. The calculation formula of the matrix is:

[0074] X∈R T×F

[0075] Where X represents the Mel spectrum feature matrix, T is the total number of time frames, and F is the number of Mel filter banks.

[0076] S4: High-fire heating and standby. When step S3 is qualified, the control unit switches the gas solenoid valve to the high-fire state to heat the boiler to the working pressure. By collecting pressure data in real time and generating Mel spectrum features, the control unit performs dynamic time warping matching with the high-fire heating benchmark C3 to monitor the boiler pressure trend. If the matching result deviates from the threshold, the gas solenoid valve is immediately closed and an alarm is issued. When the boiler reaches the set pressure, it enters the standby state and maintains a stable pressure. During this period, the pressure data is continuously monitored and matched with the standby benchmark C4. The dynamic time warping matching process includes the following steps:

[0077] T1: Mel spectrum feature matrix X∈R T×F Decompose into a time frame vector x of real-time data acquisition t , T represents the total time frames of real-time data, and x t =[x t,1 ,x t,2 ,...,x t,F ] for each x t,f Represents the energy value of the fth Mel filter output in the tth frame, where f = 1, 2, ..., F, F represents the total number of Mel filter groups; the Mel spectrum feature matrix Y∈R of the reference state obtained in advance and preprocessed U×F Decomposed into the time frame vector y of the benchmark data u , u represents the time frame index of the benchmark data, U represents the collection of time frames, where the benchmark state includes [C1, C2, C3, C4, C5];

[0078] T2: Define the local distance function. The specific calculation formula is:

[0079]

[0080] Among them, w f represents the weighting coefficient of the f-th Mel filter, and log represents the natural logarithm function;

[0081] T3: Construct a cumulative distance matrix. The specific calculation formula is:

[0082] D(t,u)=δ(x t ,y u )+min{D(t-1,u),D(t,u-1),D(t-1,u-1)}

[0083] Where D(t,u) represents the minimum cumulative matching distance between the t-th frame of real-time data and the u-th frame of reference data. The initial conditions are set as D(0,0)=0, D(t,0)=+∞, D(0,u)=+∞, where +∞ represents a sufficiently large positive number.

[0084] T4: Calculate the normalized matching distance. The specific calculation formula is:

[0085]

[0086] Where D(T,U) represents the value of the cumulative distance matrix at the last frame of real-time data and the last frame of reference data;

[0087] T5: Convert the normalized matching distance into similarity. The specific similarity calculation formula is:

[0088] Sim(X,Y)=exp(-αDist DTW (X,Y)

[0089] Among them, α is a positive definite parameter used to adjust the mapping relationship between distance and similarity.

[0090] S5: Working mode: The user turns on the steam car wash mode and uses the steam gun. The control unit periodically collects pressure data to generate Mel-spectrum features, performs dynamic time warping matching with the working benchmark C5, and determines whether there is a pressure anomaly. If the matching result is lower than the set threshold, the gas solenoid valve is quickly closed and an alarm is issued. If the matching result is normal, combustion continues to be maintained to output stable steam.

[0091] S6: Shutdown monitoring. After the user finishes washing the car, the shutdown command is executed. The control unit stops combustion and collects boiler pressure data. The obtained shutdown phase pressure time series is converted to Mel spectrum and dynamically time-warped with the shutdown benchmark C6. If it is detected that the pressure cannot drop normally or the curve deviates from the threshold during the shutdown process, an alarm is issued and the system is locked. Otherwise, the shutdown is completed after the pressure drops to a safe range, and the operating data is stored in the control unit or remote database.

[0092] The above method is implemented in a control system for preventing boiler overpressure explosions during on-site steam car washing. The system includes a boiler body, a pressure sensor, a water pump, a gas solenoid valve, an operation panel, and an electronic control unit. The specific functions of each unit are as follows:

[0093] The boiler body is used to carry and heat the incoming water flow to produce high-temperature steam;

[0094] The pressure sensor is fixedly installed on the high-pressure pipeline of the boiler body and outputs a real-time pressure signal, including an analog-to-digital converter and a buffer module;

[0095] The water pump is connected to the boiler body;

[0096] The gas solenoid valve is used to control the heating of the boiler by gas or oil, and switch between low fire and high fire according to the instructions issued by the electronic control unit;

[0097] The operation panel is used to input steam car wash mode selection and shutdown instructions and display the operating status;

[0098] The electronic control unit has a built-in algorithm module and communicates with the pressure sensor, water pump, gas solenoid valve and operation panel respectively. The algorithm module performs short-time Fourier transform and Mel scale conversion on the collected boiler pressure data to generate a Mel spectrum feature matrix, and uses dynamic time warping to perform similarity matching with the benchmark templates of each stage.

[0099] The system adopts different processing methods for different exceptions, which are specifically divided into:

[0100] When the pressure sensor output data is lost, fluctuates abnormally, or the value exceeds the preset reasonable range, the control unit automatically identifies it as a sensor failure through real-time monitoring and redundant data collection, and immediately performs backup sensor calibration, data interpolation compensation, and continuous abnormality confirmation, then closes the gas solenoid valve, activates the safety pressure relief device, and implements system shutdown operations;

[0101] When the water pump status detection result does not match the preset water supply status or detects a water supply interruption, the control unit immediately interrupts the heating operation, stops the gas supply, and starts the standby water pump detection program and alarm prompts;

[0102] When a calculation error, algorithm anomaly, or real-time matching similarity falls below a preset threshold during Mel spectrum feature conversion or dynamic time warping matching, the control unit automatically determines it as a data processing anomaly, immediately interrupts the current stage of operation, saves the abnormal data log, closes the gas solenoid valve, activates the safety pressure relief device, and issues an alarm signal;

[0103] When signal interruption, data transmission error or interference occurs in the communication interfaces between the pressure sensor, gas solenoid valve, operation panel and control unit, the system automatically starts the redundant communication channel, performs communication retry, and forces an emergency shutdown after continuous retry failures, closes the gas solenoid valve, starts the safety pressure relief device and records the fault log.

[0104] Example 2: This example uses a steam car wash device that is actually used in the market as an example. The device adopts the boiler overpressure explosion prevention control method proposed in this invention. The main technical parameters and test conditions of the device are as follows:

[0105] Boiler type: gas-heated steam boiler; maximum safe working pressure: 35 bar; normal operating pressure range: 15 to 20 bar; pressure sensor model: MPS20N0040D, sampling frequency: 2 Hz; controller model: STM32F407 microcontroller with embedded DTW matching algorithm module; steam output flow range: 3 to 20 kg / h; ambient temperature: -25°C to 35°C (simulating a low-temperature environment).

[0106] S1: System initialization and reference template loading. After the device starts, the system performs a self-test and loads the pre-saved standard reference template data. The display (HMI) prompts: "System ready, waiting for startup."

[0107] S2: During the first low-fire preheating process, the controller drives the gas valve to the low-fire state and collects pressure data in real time with a sampling frequency of 2 Hz for 30 seconds, as shown in Table 1.

[0108] Table 1: Real-time pressure data of the first low-fire preheating stage

[0109] Time(s) Pressure (bar) 0 0.2 5 0.35 10 0.55 15 0.8 20 1.05 25 1.25 30 1.4

[0110] The controller preprocesses the data (low-pass filtering and normalization) and generates a Mel spectrum matrix, as shown in Table 2.

[0111] Table 2: Mel spectrum average energy features

[0112] Frame Number Mel spectrum average energy (dB) 1 10.5 2 12.8 3 14.2 4 15.1 5 15.8

[0113] The results obtained by matching with the DTW algorithm are as follows: cumulative distance: D(T,U)=2.85; normalized matching distance: 0.285; similarity: Sim=0.87 (threshold θ=0.80); similarity 0.87>threshold 0.80, and the first warm-up is normal.

[0114] S3: The second low-fire preheating process repeats the above low-fire preheating process, as shown in Tables 3 and 4.

[0115] Table 3: Real-time pressure data of the second low-fire preheating stage

[0116] Time(s) Pressure (bar) 0 0.3 5 0.45 10 0.65 15 0.9 20 1.15 25 1.35 30 1.5

[0117] Table 4: Mel spectrum average energy features

[0118] Frame Number Mel spectrum average energy (dB) 1 11 2 13.1 3 14.6 4 15.5 5 16.2

[0119] DTW algorithm matching results: cumulative distance: D(T,U)=2.60; normalized matching distance: 0.260; similarity: Sim=0.88 (threshold θ=0.80); similarity 0.88>threshold 0.80, the second preheating is normal and enters the high-fire heating stage.

[0120] S4: During the high-fire heating stage and the standby stage, the boiler is started in high-fire heating mode with a target working pressure of 20 bar, as shown in Tables 5 and 6, and the pressure changes are monitored in real time.

[0121] Table 5: Pressure data during high-fire heating stage

[0122] Time(s) Pressure (bar) 0 1.5 10 5.2 20 8.2 30 13.0 40 20.5 50 20.0 0 1.5

[0123] Table 6: Mel spectrum features of the high-fire heating stage

[0124]

[0125]

[0126] DTW algorithm matching results: Cumulative distance: D(T,U) = 1.50; Similarity: Sim = 0.93 (threshold θ = 0.85); Similarity 0.93 > threshold 0.85, indicating normal high-fire heating. During the standby phase, the boiler automatically adjusts to low-fire mode, with pressure stabilized at 20 bar ± 0.5 bar.

[0127] S5: During the working state (actual car washing process), the user activates the steam car wash gun, and the boiler pressure fluctuates dynamically within a range of 18 to 20 bar. The system collects pressure data in real time and monitors it accordingly. Figure 2 (Simulation image) shows the real-time pressure curve during normal and abnormal operation: During normal operation, the pressure curve fluctuates smoothly (blue curve). When simulating an abnormal gun blockage, the pressure rises rapidly to 27 bar (red curve), and the DTW similarity quickly decreases to 0.45, falling below the threshold θ = 0.80, triggering automatic emergency gas valve closure and pressure relief, and the device alarm.

[0128] S6: Shutdown monitoring stage: After the user finishes washing the car, the boiler stops burning and starts to reduce the pressure, as shown in Table 7.

[0129] Table 7: Pressure change data during shutdown phase

[0130] Time(s) Pressure (bar) 0 20 10 18 20 15 30 12 40 9 50 5

[0131] DTW similarity: Sim = 0.95 (threshold θ = 0.85), shutdown is normal.

[0132] Through the detailed implementation steps, data, simulation results and quantitative analysis of the above embodiments, the present invention clearly demonstrates its advantages in boiler explosion-proof safety control, effectively improves the safety, reliability and economy of door-to-door steam car washing equipment, and has high application and promotion value.

[0133] Example 3: Reference Figure 3 This embodiment provides a specific structure and connection method of a door-to-door steam car washing boiler overpressure explosion prevention control system, which is described in detail as follows:

[0134] The system is mainly composed of a boiler body, a pressure sensor unit, a water pump unit, a gas solenoid valve unit, an operation panel unit, and an electronic control unit. The specific functions and connection relationships of each component are as follows:

[0135] Boiler body: The boiler body uses a gas-fired steam boiler with model "HG-30". The rated working pressure range is 0-35 bar, the recommended normal working range is 15-20 bar, the heating power is 8kW, and the steam generation capacity reaches 20kg / h. A high-efficiency heat exchanger is provided inside the boiler body, and a steam outlet is provided on the outside to be connected to the steam car washing gun; a water inlet is provided at the bottom of the boiler body to be connected to the water pump unit, and the boiler fuel interface is connected to the gas solenoid valve unit.

[0136] The water pump unit uses a DC booster diaphragm pump (model: DP-120), rated at 12V DC and a rated flow rate of 3L / min. One end of the pump is connected to an external water tank or source, and the other end is connected to the boiler's water inlet via high-pressure PE tubing. The pump's power supply and operating status signal lines are directly connected to the electronic control unit, which monitors the pump's operating status in real time and controls its start and stop.

[0137] The gas solenoid valve unit utilizes a controllable proportional gas valve (model: CK-258) with a rated voltage of 12V DC and a maximum operating pressure of 5.0MPa, allowing for precise control of gas flow. One end of the solenoid valve connects to an external gas cylinder or gas port via a dedicated gas hose; the other end connects to the boiler burner's gas inlet line. An electronic control unit controls the solenoid valve via PWM signals, precisely switching between low and high fire modes and shutting off the valve.

[0138] Pressure sensor unit: The pressure sensor uses the MPS20N0040D pressure sensor module with a measurement range of 0 to 40 bar, an accuracy of ±0.25%, and a response time of ≤2ms. The pressure probe is installed on the high-pressure steam outlet pipe of the boiler body. The pressure sensor unit integrates an analog-to-digital converter (ADC) and a data cache module. The ADC conversion accuracy is 12 bits and the sampling frequency is 2Hz. After ADC conversion, the pressure data is temporarily stored in the cache module and transmitted to the electronic control unit in real time through the RS485 interface for data processing.

[0139] The operation panel unit (human-machine interaction HMI) is a color LCD touch screen (model: TJC4832T035) with a screen size of 3.5 inches and a display resolution of 480×320. It has button input and status display functions. The panel communicates with the electronic control unit via a UART interface and provides the following functions:

[0140] Car wash mode selection: users can select steam car wash mode on the panel;

[0141] Start / Stop button: Users start or stop the device through the touch screen;

[0142] Operation status display: displays boiler real-time pressure, system operation status, abnormal alarm information, etc.

[0143] The electronic control unit uses the STM32F407 microcontroller as the main control chip, which has strong computing power and rich peripheral interfaces, including:

[0144] Main controller: STM32F407 microcontroller with built-in 512KB FLASH and 192KB RAM, responsible for the core tasks of the entire system, such as real-time data acquisition, signal processing, and system control logic execution;

[0145] Signal preprocessing module: filtering, denoising and normalizing the real-time data received from the pressure sensor buffer module;

[0146] Mel spectrum feature extraction module: uses short-time Fourier transform (STFT) to perform time-frequency analysis on pressure data and generates a Mel spectrum feature matrix through Mel filter bank conversion;

[0147] Dynamic Time Warping (DTW) matching algorithm module: This module performs nonlinear dynamic matching between the real-time generated Mel-spectrum feature matrix and the internally stored standard normal working state reference template, calculates the similarity, and determines the operating status of the device.

[0148] Safety control logic module: Makes safety decisions based on the real-time similarity data fed back by the DTW algorithm module. Once an abnormal pressure state is detected, it outputs a control instruction to immediately close the gas solenoid valve, stop the water pump, activate the safety pressure relief device, sound an audible and visual alarm, and notify the operator through the operation panel.

[0149] The electronic control unit interface connections are as follows: the ADC data input end is connected to the pressure sensor unit; the PWM signal output end is connected to the gas solenoid valve unit; the IO output end is connected to the water pump unit (start and stop control); the UART interface is connected to the operation panel unit; and the digital input end is connected to the water pump operation status feedback signal line.

[0150] System connection relationship: external water source → water pump unit (DP-120 booster pump) → boiler body (HG-3.0 boiler); external gas → gas solenoid valve unit (CK-258 solenoid valve) → boiler body burner; boiler body high-pressure pipeline → pressure sensor unit (MPS20N0040D) → electronic control unit ADC input; operation panel (TJC4832T035) → UART communication → electronic control unit; electronic control unit → PWM output → gas solenoid valve (controls low fire, high fire and valve opening and closing); electronic control unit → IO output → water pump (controls start and stop); water pump operation status signal → electronic control unit (IO input); electronic control unit → control logic module → operation panel (displays status and alarm information).

[0151] The system constructed by the above-mentioned embodiment can accurately monitor the boiler pressure change trend and abnormal conditions in real time through the dual low-fire preheating discrimination process, Mel spectrum feature analysis and DTW matching algorithm. When the system pressure sensor is frozen in winter (-5°C) or the sensor fails, the reaction time is significantly shortened compared to the traditional system, and the success rate of abnormality detection is significantly improved. Under the conditions of multiple on-site experiments, the pressure anomaly recognition rate of the equipment in this embodiment is as high as over 98%, and the system safety response time is within 1-2 seconds, which effectively reduces the risk of overpressure explosion of the boiler of the on-site steam car washing equipment and improves the safety of users and the reliability of stable operation of the equipment.

[0152] 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 control method for preventing boiler overpressure explosion during door-to-door steam car washing, characterized in that: The method comprises the following steps: S1: Initialize the device, start the control unit, load the Mel spectrum reference template and dynamic time warping matching algorithm parameters for each stage, and perform communication and hardware self-test on the pressure sensor, water pump, and gas solenoid valve; S2: First low-fire preheating judgment: The control unit drives the gas solenoid valve to the low-fire state to preheat the boiler and collects pressure data in real time. The collected pressure data is preprocessed to obtain a pressure time series. Short-time Fourier transform and Mel scaling conversion are used to generate Mel spectrum features. This feature matrix is matched with the first preheating benchmark C1 through dynamic time warping for similarity. If the similarity result meets the threshold, the next step is entered; otherwise, heating is stopped and an alarm is issued. S3: Second low-fire preheating determination: Repeat the acquisition and Mel spectrum feature extraction process in step S2. Use dynamic time warping to calculate the distance between the obtained second preheating Mel spectrum feature and the second preheating benchmark C2 and obtain the similarity. If the similarity result meets the threshold, subsequent high-fire heating is allowed. Otherwise, heating is stopped and an alarm is issued. S4: High-fire heating and standby. If step S3 is qualified, the control unit switches the gas solenoid valve to the high-fire state to heat the boiler to the working pressure. By collecting pressure data in real time and generating Mel-spectrum features, the control unit performs dynamic time warping matching with the high-fire heating benchmark C3 to monitor the boiler pressure trend. If the matching result deviates from the threshold, the gas solenoid valve is immediately closed and an alarm is issued. When the boiler reaches the set pressure, it enters the standby state and maintains a stable pressure. During this period, the pressure data is continuously monitored and matched with the standby benchmark C4. S5: Working mode: The user turns on the steam car wash mode and uses the steam gun. The control unit periodically collects pressure data to generate Mel-spectrum features, performs dynamic time warping matching with the working benchmark C5, and determines whether there is a pressure anomaly. If the matching result is lower than the set threshold, the gas solenoid valve is quickly closed and an alarm is issued. If the matching result is normal, combustion continues to be maintained to output stable steam. S6: Shutdown monitoring. After the user finishes washing the car, the shutdown command is executed. The control unit stops combustion and collects boiler pressure data. The obtained shutdown phase pressure time series is converted to Mel spectrum and dynamically time-warped with the shutdown benchmark C6. If it is detected that the pressure cannot drop normally or the curve deviates from the threshold during the shutdown process, an alarm is issued and the system is locked. Otherwise, the shutdown is completed after the pressure drops to a safe range, and the operating data is stored in the control unit or remote database.

2. The control method for preventing boiler overpressure explosion during door-to-door steam car washing according to claim 1, characterized in that: The pressure data acquisition process described in steps S2-S5 includes the following steps: first, the pressure signal is continuously acquired from the pressure sensor through an analog-to-digital converter at a fixed sampling frequency, and then the acquired original signal is filtered, denoised and normalized and stored in the cache module of the control unit according to the time frame, and finally the pressure data in the cache is transmitted to the algorithm module for Mel spectrum conversion and dynamic time warping matching.

3. The control method for preventing boiler overpressure explosion during door-to-door steam car washing according to claim 1, characterized in that: The method for generating the Mel spectrum features in steps S2-S5 includes the following steps: First, the pressure time series after filtering and normalization is framed and windowed, and short-time Fourier transform is performed to obtain the time-frequency amplitude information P(t,ω). The calculation formula of short-time Fourier transform is: Where p(τ) represents the value of the pressure signal at time τ, w(·) represents the window function, N is the number of sampling points in each frame, the window function w(t-τ) is used to reduce the frame edge effect, t represents the time frame index, P(t,ω) represents the complex amplitude at the t-th frame and angular frequency ω, and j is the imaginary unit; Then, we take the modulus of P(t,ω) to get the amplitude spectrum, and then map the linear frequency f to the Mel frequency M(f). The calculation formula of this association is: Where M(f) represents the corresponding Mel scale frequency value; Then, the energy of P(t,ω) is integrated on the Mel scale using a pre-set Mel filter bank to obtain the energy vector of each time frame. Finally, logarithmic compression is performed on the energy vector to generate a two-dimensional Mel spectrum feature matrix. The calculation formula of the matrix is: X∈R T×F Where X represents the Mel spectrum feature matrix, T is the total number of time frames, and F is the number of Mel filter banks.

4. The control method for preventing boiler overpressure explosion during door-to-door steam car washing according to claim 1, characterized in that: The dynamic time warping matching process The following steps are involved: T1: Mel spectrum feature matrix X∈R T×F Decompose into a time frame vector x of real-time data acquisition t , T represents the total time frames of real-time data, and x t =[x t,1 ,x t,2 ,...,x t,F ] for each x t,f Represents the energy value of the fth Mel filter output in the tth frame, where f = 1, 2, ..., F, F represents the total number of Mel filter groups; the Mel spectrum feature matrix Y∈R of the reference state obtained in advance and preprocessed U×F Decomposed into the time frame vector y of the benchmark data u , u represents the time frame index of the benchmark data, U represents the collection of time frames, where the benchmark state includes [C1, C2, C3, C4, C5]; T2: Define the local distance function. The specific calculation formula is: Among them, w f represents the weighting coefficient of the f-th Mel filter, and log represents the natural logarithm function; T3: Construct a cumulative distance matrix. The specific calculation formula is: D(t,u)=δ(x t ,y u )+min{D(t-1,u),D(t,u-1),D(t-1,u-1)} Where D(t,u) represents the minimum cumulative matching distance between the t-th frame of real-time data and the u-th frame of reference data. The initial conditions are set as D(0,0)=0, D(t,0)=+∞, D(0,u)=+∞, where +∞ represents a sufficiently large positive number. T4: Calculate the normalized matching distance. The specific calculation formula is: Where D(T,U) represents the value of the cumulative distance matrix at the last frame of real-time data and the last frame of reference data; T5: Convert the normalized matching distance into similarity. The specific similarity calculation formula is: Sim(X,Y)=exp(-αDist DTW (X,Y)) Among them, α is a positive definite parameter used to adjust the mapping relationship between distance and similarity.

5. A control system for preventing boiler overpressure explosion during on-site steam car washing, characterized by: The control system is used to implement the control method according to any one of claims 1 to 4. The system includes a boiler body, a pressure sensor, a water pump, a gas solenoid valve, an operation panel and an electronic control unit. The specific functions of each unit are: The boiler body is used to carry and heat the incoming water flow to generate high-temperature steam; The pressure sensor is fixedly installed on the high-pressure pipeline of the boiler body and outputs a real-time pressure signal, including an analog-to-digital converter and a buffer module; The water pump is connected to the boiler body; The gas solenoid valve is used to control the heating of the boiler by gas or oil, and switches between low fire and high fire according to the instructions issued by the electronic control unit; The operation panel is used to input steam car wash mode selection and shutdown instructions and display the operating status; The electronic control unit has a built-in algorithm module and is respectively connected to the pressure sensor, water pump, gas solenoid valve and operation panel. The algorithm module performs short-time Fourier transform and Mel scale conversion on the collected boiler pressure data to generate a Mel spectrum feature matrix, and uses dynamic time warping to perform similarity matching with the benchmark templates of each stage.

6. A control system for preventing boiler overpressure explosion during on-site steam car washing according to claim 5, characterized in that: The system adopts different processing methods for different exceptions, which are specifically divided into: When the pressure sensor output data is lost, fluctuates abnormally, or the value exceeds the preset reasonable range, the control unit automatically identifies it as a sensor failure through real-time monitoring and redundant data collection, and immediately performs backup sensor calibration, data interpolation compensation, and continuous abnormality confirmation, then closes the gas solenoid valve, activates the safety pressure relief device, and implements system shutdown operations; When the water pump status detection result does not match the preset water supply status or detects a water supply interruption, the control unit immediately interrupts the heating operation, stops the gas supply, and starts the standby water pump detection program and alarm prompts; When a calculation error, algorithm anomaly, or real-time matching similarity falls below a preset threshold during Mel spectrum feature conversion or dynamic time warping matching, the control unit automatically determines it as a data processing anomaly, immediately interrupts the current stage of operation, saves the abnormal data log, closes the gas solenoid valve, activates the safety pressure relief device, and issues an alarm signal; When signal interruption, data transmission error or interference occurs in the communication interfaces between the pressure sensor, gas solenoid valve, operation panel and control unit, the system automatically starts the redundant communication channel, performs communication retry, and forces an emergency shutdown after continuous retry failures, closes the gas solenoid valve, starts the safety pressure relief device and records the fault log.

Images