Containerized emergency multifunctional repair shelter for remote site

Abstract

The application discloses a container type remote site emergency multifunctional repair shelter, which comprises a shelter cabin body, an electric arc additive equipment, a portable milling machine and a laser additive equipment which are matched and installed in the shelter cabin body, the electric arc additive equipment comprises an electric arc additive device, a multi-source signal acquisition module, a molten drop transition mode identification module and a process parameter self-adaptive adjustment module, the multi-source signal acquisition module is used for collecting environmental vibration signals of a position where a welding torch at the tail end of a welding additive execution head and electric parameter signals output by a welding power source in real time, the molten drop transition state identification module is used for identifying a current molten drop transition mode according to the vibration signals and the electric parameter signals, and the process parameter self-adaptive adjustment module is used for dynamically adjusting welding process parameters, and the application realizes molten drop transition self-adaptive stability through vibration and molten pool coupling closed-loop control, and effectively improves forming quality and repair reliability in a vibration environment.

Description

Technical Field

[0001] This invention belongs to the field of remote site repair and manufacturing technology, specifically relating to a containerized remote site emergency multi-functional repair cabin. Background Technology

[0002] A field repair module is a mobile emergency repair platform integrated into a standard shipping container. It typically carries arc additive manufacturing (AIM) equipment and is used for rapid metal deposition and forming repairs of worn or broken parts of heavy equipment in remote areas or battlefield environments. AIM uses welding wire as raw material, melting it with an electric arc and depositing it layer by layer to form a metal structure. It boasts advantages such as high deposition efficiency, high material utilization, and low equipment cost, making it a core technology for module-level on-site repair.

[0003] Currently, research on mobile cabin-level arc additive manufacturing technology mainly focuses on integrating various additive manufacturing equipment, 3D scanning devices, and vibration damping and isolation devices into mobile cabins to solve the problems of equipment transportation fixation and spatial adaptability. In terms of vibration control, existing solutions generally adopt passive vibration isolation measures (such as rubber pads, spring dampers, or integrated vibration damping and isolation devices) to attempt to isolate environmental vibrations from the additive manufacturing execution system. However, the actual operating environment of field cabins is characterized by multi-source complex vibrations, including low-frequency impacts (0.5~20Hz) induced by road transportation, medium-frequency continuous vibrations (50~100Hz) generated by generator operation, dynamic excitation (10~50Hz) brought about by robotic arm movement, and high-frequency oscillations of the arc itself. These vibrations are transmitted through the cabin floor to the substrate and the end of the welding torch, directly changing the force balance of the molten droplet detaching from the welding wire. This causes the droplet transfer mode to drift from a stable jet transfer to a short-circuit transfer or an irregular multi-droplet state, resulting in uncontrolled formation consistency of the cladding layer and frequent occurrences of porosity and incomplete fusion defects. In addition, the oxidation level of the welding wire surface changes dynamically in high humidity environments in the field, while existing cleaning devices use fixed operating parameters and cannot adjust the cleaning intensity in real time according to the oxidation level. Furthermore, they do not incorporate the surface condition of the welding wire and the stability of the droplet transition into unified control. Therefore, the existing modular electric arc additive manufacturing equipment has a low repair success rate and a high rework rate in multi-source vibration environments, making it difficult to meet the requirements for rapid emergency repairs in the field. Summary of the Invention

[0004] To address the shortcomings and problems of existing repair cabins, this invention provides a containerized remote site emergency multifunctional repair cabin that can achieve adaptive stability of droplet transition through vibration and molten pool coupled closed-loop control, effectively improving the forming quality and repair reliability under vibration environment.

[0005] The solution adopted by this invention to solve its technical problem is: a containerized remote site emergency multi-functional repair cabin, including a cabin body, in which an arc additive manufacturing equipment, a portable milling machine, and a laser additive manufacturing equipment are matched and installed. The arc additive manufacturing equipment includes an arc additive manufacturing device, a multi-source signal acquisition module, a droplet transition mode recognition module, and a process parameter adaptive adjustment module. The multi-source signal acquisition module is installed on the welding additive manufacturing execution head of the arc additive manufacturing device and is used to collect in real time the environmental vibration signal at the location of the welding torch at the end of the welding additive manufacturing execution head and the electrical parameter signal output by the welding power source. The droplet transition state recognition module is used to identify the current droplet transition mode based on the vibration signal and the electrical parameter signal. The process parameter adaptive adjustment module is connected to the arc additive controller of the arc additive manufacturing device and is used to dynamically adjust the welding process parameters based on the difference between the identified current droplet transition mode and the preset target stable transition mode.

[0006] The multi-source signal acquisition module includes a vibration sensing unit and an electrical parameter sampling unit. The vibration sensing unit is installed on the welding torch holder of the welding additive manufacturing head and is used to acquire vibration acceleration signals and / or angular velocity signals in three-dimensional space. The electrical parameter sampling unit is used to synchronously acquire welding current and / or arc voltage signals.

[0007] The droplet transition mode recognition module includes a signal preprocessing unit and a mode classification unit. The signal preprocessing unit is connected to the multi-source signal acquisition module and is used to receive the vibration signal and electrical parameter signal acquired by the signal preprocessing unit, and to perform noise reduction and feature extraction on the vibration signal and electrical parameter signal. The mode classification unit is used to determine the current droplet transition mode type based on the extracted feature vector and a classification model.

[0008] The signal preprocessing unit performs time-frequency domain transformation on the vibration signal to extract the energy distribution characteristics within a predetermined frequency band; the predetermined frequency band includes at least one of the low-frequency impact band, the mid-frequency continuous excitation band, and the high-frequency oscillation band.

[0009] The classification model is a neural network-based classifier. Its input features include at least the frequency band energy features and electrical parameter waveform features of the vibration signal, and its output is the probability or category of short-circuit transition, droplet transition or jet transition.

[0010] The adaptive adjustment module for process parameters includes an arc parameter adjustment unit connected to the arc additive manufacturing controller. The arc parameter adjustment unit is used to dynamically adjust the welding parameters according to the identified differences. The welding parameters include at least one of the following: peak pulse current, base current, pulse duty cycle, welding voltage, or wire feed speed.

[0011] The process parameter adaptive adjustment module also includes a motion parameter adjustment unit, which is used to dynamically adjust the movement speed and / or path overlap rate of the robotic arm according to the intensity or frequency band characteristics of the vibration signal.

[0012] The multi-source signal acquisition module also includes a welding wire surface condition detection unit, which is used to detect the oxidation degree of the welding wire surface online and send the detection result to the process parameter adaptive adjustment module; the process parameter adaptive adjustment module also includes a cleaning parameter adjustment unit connected to the arc additive manufacturing controller, which sends cleaning parameters to the arc additive manufacturing controller according to the detection result of the oxidation degree of the welding wire surface.

[0013] The beneficial effects of this invention are as follows: The containerized remote emergency multi-functional repair cabin provided by this invention effectively increases the proportion of time that the droplet transition mode is maintained in the stable droplet transition range, reduces the droplet detachment frequency fluctuation rate, and reduces the on-site rework rate caused by vibration by synchronously acquiring multi-source signals of vibration and electrical parameters, extracting wavelet packet frequency band energy features, and classifying droplet transition modes using BP neural networks, combined with improved incremental PID arc parameter adjustment and coordinated control of robotic arm motion parameters. At the same time, the closed-loop control of welding wire cleaning intensity based on laser reflectivity R effectively reduces the spatter rate caused by welding wire oxidation. The end-to-end control delay of parallel acquisition and decision-making by multi-core heterogeneous processor is less than 3ms, and there is no need to install a large vibration isolation platform in the cabin. Adaptive control can be achieved only through a micro-sensor unit integrated at the end of the welding torch and an embedded control board. Attached Figure Description

[0014] Figure 1 This is a three-dimensional structural schematic diagram of the present invention.

[0015] Figure 2 This is a schematic diagram of the connection frame of each module of the electric arc additive manufacturing equipment of the present invention.

[0016] Figure 3 This is a flowchart of wavelet packet denoising and feature extraction of the present invention.

[0017] Figure 4 This is a block diagram of the closed-loop control of welding wire cleaning strength in this invention.

[0018] Figure 5 This is a timing diagram of the swimlane for the collaborative operation of the hardware and software in this invention.

[0019] The diagram is labeled as follows: 1 is the container body, 11 is the additive manufacturing compartment, 12 is the milling machine compartment, 2 is the electric arc additive manufacturing equipment, and 3 is the portable milling machine. Detailed Implementation

[0020] The present invention will be further described below with reference to the accompanying drawings and embodiments. Example

[0021] This embodiment provides a containerized remote site emergency multi-functional repair cabin, such as... Figure 1 As shown, the container includes a modular cabin 1, with safety doors on one or more sides for entering the interior. The cabin contains a power compartment, an air supply compartment, a tool and consumables compartment, and an equipment compartment, all separated by fireproof partitions. The equipment compartment is divided by fireproof partitions into an additive manufacturing compartment 11 and a milling machine compartment 12. The additive manufacturing compartment is equipped with additive manufacturing equipment, including laser additive manufacturing equipment and an electric arc additive manufacturing equipment 2. The milling machine compartment 12 contains a portable milling machine 3. The electric arc additive manufacturing equipment includes an electric arc additive device, a multi-source signal acquisition module, a droplet transition mode recognition module, a process parameter adaptive adjustment module, and a multi-core heterogeneous processing platform. The multi-source signal acquisition module, droplet transition mode recognition module, and process parameter adaptive adjustment module are all integrated into or communicatively connected to the multi-core heterogeneous processor.

[0022] In this embodiment, the arc additive manufacturing device is an existing arc additive welding device, including a six-axis industrial robotic arm, a welding additive execution head mounted on the flange at the end of the robotic arm, a welding wire supply mechanism, a welding power source, a welding wire cleaning mechanism, and an arc additive controller. The arc additive controller is used to control the operation of the entire arc additive welding device according to parameters.

[0023] A multi-source signal acquisition module is installed on the welding additive manufacturing head. Its output is connected to the input of the droplet transition state recognition module via a multi-core heterogeneous processor. The multi-core heterogeneous processor is used to acquire, in real time, the environmental vibration signal at the end of the welding torch and the electrical parameter signal output by the welding power source during the arc additive manufacturing process, and sends the acquired signals to the droplet transition state recognition module. Specifically: The multi-source signal acquisition module includes a vibration sensing unit and an electrical parameter sampling unit. A metal shielding shell is fixedly installed on the welding torch holder of the welding additive manufacturing head. The vibration sensing unit is fixedly installed inside the metal shielding shell. The digital signal output by the vibration sensing unit is connected to the multi-core heterogeneous processor via an I²C bus. The sensing center of the vibration sensing unit is less than 100mm away from the center of the molten pool arc column. The vibration sensing unit is used to acquire vibration acceleration and angular velocity signals in three-dimensional space, and sends the acquired vibration acceleration and angular velocity signals to the droplet transition state identification module through the multi-core heterogeneous processor. In this embodiment, the sampling frequency of the vibration sensing unit is set to 10kHz. The vibration sensing unit uses a six-axis MEMS chip of model MPU-6050, which includes a three-axis accelerometer and a three-axis gyroscope. The metal shielding shell is made of nickel-plated copper to reduce electromagnetic interference during welding. The signal line of the vibration sensing unit is a twisted-pair shielded cable, and a ferrite magnetic ring is wrapped near the connector.

[0024] In this embodiment, the SCL and SDA pins of the vibration sensing unit chip are connected to the input of the digital isolator ADuM1250 after passing through a bidirectional level conversion circuit (3.3V to 5V). The output of the isolator is connected to the I²C controller of the processor. The power supplies on both sides of the isolator (VDD1 and VDD2) are powered by independent DC-DC modules, forming a complete electrical isolation barrier.

[0025] The electrical parameter sampling unit is connected to a multi-core heterogeneous processor to synchronously acquire welding current and arc voltage signal data. The acquired welding current and arc voltage signal data are then sent to the droplet transfer state identification module via the multi-core heterogeneous processor. There are various methods for acquiring welding current and arc voltage signals, such as: Figure 4 As shown, in this embodiment, the welding current is acquired by a clamp-on Hall effect sensor (model ACS758, range ±200A, bandwidth 120kHz); the arc voltage is directly taken from the terminal voltage between the welding torch and the substrate of the welding additive manufacturing head through a differential attenuation circuit; the start signal of the electrical parameter sampling unit is triggered by the timer of the multi-core heterogeneous processor to ensure that it is aligned with the sampling time of the vibration sensor, and the sampling clock deviation is controlled within ±1μs.

[0026] The droplet transition state identification module includes a signal preprocessing unit and a pattern classification unit. The signal preprocessing unit is connected to a multi-core heterogeneous processor and is used to receive signals acquired by the multi-source signal acquisition module, perform noise reduction processing on the acquired vibration signals and electrical parameter signals, and extract features from the noise-reduced signal data. Specifically: In this embodiment, the signal preprocessing unit uses a second-order active low-pass filter to reduce noise in the welding current signal and arc voltage signal collected by the electrical parameter sampling unit; for example... Figure 3 As shown, the signal preprocessing unit in this embodiment uses wavelet packet transform to perform adaptive noise reduction on the vibration signal. The specific adaptive noise reduction of the vibration signal includes the following steps: Step 1: Perform 4-level wavelet packet decomposition on the vibration signal sequence x(n) of length 2048 points, selecting db8 as the wavelet basis, to obtain 2 4 =16 frequency band sub-sequences.

[0027] Step 2: Soft thresholding is applied to the coefficients of each subsequence. The threshold is adaptively calculated using the Stein unbiased risk estimation method.

[0028] Step 3: Reassemble the processed coefficients to obtain the noise-reduced vibration signal x'(n).

[0029] Compared to traditional low-pass filtering, wavelet packet denoising can preserve transient impact components in the signal (such as high-frequency spikes caused by road impacts), which are crucial for identifying low-frequency interference.

[0030] Feature extraction includes performing time-frequency domain transformation on the vibration signal to extract energy distribution features within a predetermined frequency band. In this embodiment, the predetermined frequency band includes the low-frequency impact band, the mid-frequency continuous excitation band, and the high-frequency oscillation band. The signal preprocessing unit estimates the power spectral density of the noise-reduced vibration signal (using the Welch method, with a Hanning window function, a window length of 256 points, and an overlap rate of 50%), and calculates the energy proportion within the low-frequency impact band, the mid-frequency continuous excitation band, and the high-frequency oscillation band. Specifically: The low-frequency impact band is 0.5-10Hz, mainly reflecting the road impact and vehicle shaking of the container base plate; the medium-frequency continuous excitation band is 10-200Hz, including the generator's 50-100Hz fundamental frequency and its harmonics and the vibration caused by the low-speed movement of the robotic arm; the high-frequency oscillation band is 200-1000Hz, reflecting the resonance of the welding torch end structure and the fluctuation of the arc pressure.

[0031] Let the total energy be E_total, and the energies of each frequency band be E_low, E_mid, and E_high. Then the extracted feature vector is [E_low / E_total, E_mid / E_total, E_high / E_total, E_low, E_mid, E_high], which has a total of 6 dimensions. For electrical parameter signals, this embodiment uses a time-domain feature extraction method: calculate the peak current Ip, the rising slope di / dt (taking the linear fitting slope of the rising segment from 10% to 90%), and the pulse width tp from the current waveform within each pulse period; calculate the peak-to-peak fluctuation amplitude Vpp within one period from the voltage waveform; use the four time-domain features of the electrical parameter signal as auxiliary features, and combine them with the six frequency band energy features of the vibration signal to form a 10-dimensional feature vector.

[0032] The pattern classification unit uses the extracted feature vector to determine the current droplet transition mode type using a classification model. In this embodiment, the classification model is a neural network-based classifier. Its input features include at least the frequency band energy features and electrical parameter waveform features of the vibration signal, and the output is the probability or category of short-circuit transition, droplet transition or jet transition.

[0033] In this embodiment, the classification model uses a three-layer BP neural network to classify droplet transition modes. The network structure is as follows: Input layer: 10 nodes, corresponding to six vibration frequency band energy characteristics and four electrical parameter time-domain characteristics. The six vibration frequency band energy characteristics are: the energy ratio of the low-frequency impact band E_low / E_total, the energy ratio of the mid-frequency continuous excitation band E_mid / E_total, the energy ratio of the high-frequency oscillation band E_high / E_total, and the absolute energy values ​​of the three frequency bands E_low, E_mid, and E_high. The four electrical parameter time-domain characteristics are: peak current Ip, current rise slope di / dt, pulse width tp, and voltage fluctuation amplitude Vpp.

[0034] Hidden layer: The number of nodes was determined to be 12 through trial and error, and the activation function was the sigmoid function f(x)=1 / (1+e^{-x}); Output layer: 3 nodes, corresponding to the three modes of short-circuit transition, droplet transition and jet transition respectively. The activation function is softmax, and the output value represents the probability of belonging to each category.

[0035] In this embodiment, the training of the three-layer BP neural network was completed offline. The training data was collected from standard welding tests conducted on a laboratory simulated vibration table (programmable vibration spectrum, including sinusoidal sweep frequencies of 0.5-200Hz and random impacts). At least 2000 samples were collected for each transition mode, and the data were divided into training and validation sets at a ratio of 3:1. The stochastic gradient descent algorithm was used, with an initial learning rate of 0.01, a momentum factor of 0.9, and a loss function of cross-entropy. The training run consisted of 1000 epochs, and the validation accuracy reached 94.8%. After training, the network weights and bias parameters were stored in the processor's internal Flash memory.

[0036] During online inference, the multi-core heterogeneous processor performs a forward computation every 50ms (corresponding to 500 sampling points): the 10-dimensional feature vector extracted in real time is input into the network, the output probability is calculated layer by layer, the mode corresponding to the maximum probability is selected as the judgment result of the current droplet transition mode, and the confidence score is output at the same time.

[0037] Furthermore, when the maximum probability value of the neural network output is less than 0.6 (i.e., the confidence level is low), a hard decision is made in conjunction with the arc voltage fluctuation amplitude Vpp. If Vpp > 5V and periodic short-circuit spikes appear, it is determined to be a short-circuit transition; if Vpp is between 2 and 4V and the waveform is smooth, it is determined to be a droplet transition; if Vpp < 2V and the current is in a continuous jetting pattern, it is determined to be a jet transition, thus preventing misjudgment in extreme noise environments.

[0038] The control signal output terminal of the real-time adaptive adjustment module for process parameters is connected to the communication interface of the arc additive manufacturing controller. The adaptive adjustment module for process parameters includes an arc parameter adjustment unit, which is used to dynamically adjust at least one of the following welding parameters: peak pulse current, pulse duty cycle, welding voltage, or wire feed speed.

[0039] For example, when the droplet transition state identification module detects that the current droplet transition mode deviates from the target mode (the default target mode in this embodiment is droplet ejection transition), the incremental PID control algorithm is activated. The controlled object is the peak pulse current Ip, and the controlled variable setpoint Ip_ref corresponds to the typical value of the droplet ejection transition stable range (for a 1.2mm diameter ER70S-6 welding wire, Ip_ref=280A). The control deviation e(k)=Ip_ref-Ip(k), where k is the discrete sampling time (updated every 50ms).

[0040] The incremental PID control law is: Δu(k)=Kp·[e(k)-e(k-1)]+Ki·e(k)+Kd·[e(k)-2e(k-1)+e(k-2)] The actual output control quantity is u(k) = u(k-1) + Δu(k). The parameters are tuned on-site: Kp = 0.6, Ki = 0.08, Kd = 0.15. After being limited (200A~350A), the output control quantity is sent to the arc additive manufacturing controller via a PWM signal (20kHz carrier) through optocoupler isolation. The arc additive manufacturing controller adjusts the welding power supply current according to the output control quantity.

[0041] In addition to the peak pulse current, when the deviation persists for more than 3 sampling cycles (i.e., it has not returned after 150ms), the process parameter adaptive adjustment module simultaneously adjusts the pulse duty cycle (gradually increasing from 35% to 45%) to enhance the control strength.

[0042] Furthermore, the adaptive process parameter adjustment module also includes a motion parameter adjustment unit connected to the arc additive controller, used to dynamically adjust the robot arm's movement speed and / or path overlap rate based on the intensity or frequency band characteristics of the vibration signal via the arc additive controller. Specifically: When the mid-frequency energy ratio (E_mid / E_total) of the vibration signal exceeds the threshold of 0.35 (indicating strong vibration of the generator or robotic arm) or the low-frequency energy ratio exceeds 0.25 (indicating road impact), the system determines that the current operating condition is causing significant disturbance to the molten pool, and it is necessary to reduce the robotic arm's movement speed to decrease the additional inertial force. The adjustment logic is as follows: If only low-frequency impacts are significant (E_low / E_total>0.25), the moving speed decreases linearly from the baseline value v0 to 0.6·v0, and the path overlap rate increases from 50% to 65%; where v0 is the standard moving speed set in the current program (e.g., 8 mm / s).

[0043] If only the mid-frequency excitation is significant (E_mid / E_total>0.35), the velocity decreases to 0.7·v0, and the overlap rate increases to 60%.

[0044] If both are significant, the velocity decreases to 0.5·v0, and the overlap rate increases to 70%.

[0045] Speed ​​adjustment commands are sent to the robotic arm controller via the EtherCAT bus. The controller performs local trajectory replanning in joint space—re-interpolating only the latter half of the current motion segment to avoid computational delays caused by global replanning. Motion parameter adjustments and arc parameter adjustments are performed in parallel, without the need for strict time synchronization between them.

[0046] This embodiment uses the RK3588 multi-core heterogeneous processor as the main control chip, which integrates four Cortex-A76 big cores (running Linux), four Cortex-A55 little cores, one Cortex-M0 coprocessor, and one NPU with 3 TOPS computing power. There are various resource allocation methods for multi-core heterogeneous processors, such as: Cortex-M0 core: Bare-metal programming, responsible for 10kHz hard real-time data acquisition: periodically triggering ADC conversion, reading vibration data and CMOS image data from the I²C bus, packaging the data into shared memory (SRAM partition) and updating the flag bits. This core is also responsible for generating PWM control signals.

[0047] NPU + one Cortex-A76 core: Deploys a trained BP neural network model to accelerate forward inference and writes the inference results to shared memory.

[0048] Another Cortex-A76 core: runs the process parameter decision program (PID controller and cleanup intensity controller), reads the identification results from shared memory, calculates the control increment, and updates the PWM duty cycle and EtherCAT instructions.

[0049] The remaining Cortex-A76 cores are used for running the human-machine interface (displaying real-time vibration spectrum, droplet transfer mode, welding parameters, etc. via a 7-inch touchscreen) and communication management (exchanging status data with the control center of the mobile cabin).

[0050] Inter-core communication employs shared memory in conjunction with a spinlock mechanism. Key data (such as the current droplet transition mode and PID control increments) is updated using a double-buffered approach to ensure that real-time data is not overwritten. Actual measurements show that the end-to-end latency from sensor sampling to control quantity update is less than 3ms (sampling period 50ms, neural network inference time approximately 1.2ms, PID calculation <0.1ms, inter-core data copy <0.5ms, with the remainder being scheduling wait). Example

[0051] The difference between Example 2 and Example 1 is that the multi-source signal acquisition module further includes a welding wire surface condition detection unit, which is used to detect the oxidation degree of the welding wire surface online and send the detection result to the process parameter adaptive adjustment module. The process parameter adaptive adjustment module also includes a cleaning parameter adjustment unit connected to the arc additive manufacturing controller. The cleaning parameter adjustment unit sends cleaning parameters to the arc additive manufacturing controller according to the detection result of the oxidation degree of the welding wire surface. The arc additive manufacturing controller adaptively adjusts the cleaning intensity of the welding wire cleaning mechanism according to the cleaning parameter signal to control the surface condition of the welding wire entering the molten pool to be in the preset optimal range.

[0052] The welding wire surface condition detection unit uses laser reflection to detect the oxidation degree of the welding wire surface in real time. The welding wire surface condition detection unit includes a line laser module and a CMOS linear array image sensor connected to a multi-core heterogeneous processor. The line laser module is arranged on both sides of the wire feeding tube in the wire feeding system at a 45° incident angle. The laser from the line laser module is incident on the welding wire surface at a 45° angle. The reflected light is focused onto the photosensitive surface of the CMOS sensor by a cylindrical lens. The multi-core heterogeneous processor controls the activation of the laser and the exposure and readout of the CMOS sensor. The multi-core heterogeneous processor collects one frame of reflected light intensity distribution data every 20ms. By calculating the ratio R = Ir / I0 of the peak intensity of the reflected light to the factory-calibrated incident light intensity I0 of the laser, and determining the oxidation degree range of R, the oxidation degree of the welding wire surface is characterized: R > 0.85 indicates that the welding wire is clean (extremely thin oxide film), R < 0.30 indicates severe oxidation, and the range between the two is normal. This data is transmitted to the process parameter adaptive adjustment module via the shared memory between the cores of the multi-core heterogeneous processor. The surface oxidation degree detection unit of the welding wire outputs the reflectivity ratio R in real time. When R is lower than the first threshold of 0.40, it indicates that the oxide film on the surface of the welding wire is too thick and the cleaning intensity needs to be increased. When R is higher than the second threshold of 0.75, it indicates that the welding wire is too clean (which may reduce the stability of droplet transfer), and the minimum cleaning power should be maintained.

[0053] In this embodiment, the welding wire cleaning mechanism consists of an annular high-pressure airflow nozzle and a rotating brush. The airflow pressure is controlled by a pneumatic proportional valve (continuously adjustable from 0 to 0.6 MPa), and the brush rotation speed is driven by a small servo motor (0 to 3000 rpm). The control algorithm uses a single-input single-output PID controller with feedback R and a setpoint of R_set = 0.55 (the center value of the optimal oxidation range). The PID output is mapped to the proportional valve opening and the motor speed. Proportional valve opening = base opening (0.2MPa corresponds to 20% opening) + 0.8 × Δu' Motor speed = base speed (500 rpm) + 2000 × Δu' Where Δu' is the normalized PID output (-1 to 1). When R rises back to the 0.5 to 0.65 range and remains there for more than 1 second, the cleaning intensity gradually returns to the baseline value.

[0054] It should be noted that the above embodiments and accompanying drawings are merely illustrative examples of the core principles and key structures of the "Containerized Remote Site Emergency Multifunctional Repair Cabin" of the present invention. The accompanying drawings are simplified schematic diagrams, intended to clearly illustrate the structural, process, or data flow relationships related to the innovative points of the technical solution, and are not intended to limit the complete form of the actual product. This specification focuses on the innovative technical means necessary to achieve the invention's objectives and solve the technical problems. While auxiliary or common-sense details such as "dustproof design," "heat dissipation layout," "interface protocol," "conventional filtering," and "standard component selection," which can be implemented without creative effort by those skilled in the art, are not elaborated upon, they should be understood as naturally encompassed in the specific implementation of the present invention and fall within the protection and implementation scope of this technical solution. The scope of protection of the present invention is determined by the claims, and the specification and accompanying drawings are only used to interpret the claims.

Claims

1. A containerized remote site emergency multi-functional repair cabin, comprising a cabin body, wherein an arc additive manufacturing device, a portable milling machine, and a laser additive manufacturing device are fitted and installed within the cabin body, characterized in that, The arc additive manufacturing equipment includes an arc additive manufacturing device, a multi-source signal acquisition module, a droplet transfer mode recognition module, and a process parameter adaptive adjustment module. The multi-source signal acquisition module is installed on the welding additive manufacturing execution head of the arc additive manufacturing device and is used to acquire in real time the environmental vibration signal at the location of the welding torch at the end of the welding additive manufacturing execution head and the electrical parameter signal output by the welding power source. The droplet transfer state recognition module is used to identify the current droplet transfer mode based on the vibration signal and the electrical parameter signal. The process parameter adaptive adjustment module is connected to the arc additive controller of the arc additive manufacturing device and is used to dynamically adjust the welding process parameters based on the difference between the identified current droplet transfer mode and the preset target stable transfer mode.

2. The containerized remote site emergency multi-functional repair cabin according to claim 1, characterized in that, The multi-source signal acquisition module includes a vibration sensing unit and an electrical parameter sampling unit. The vibration sensing unit is installed on the welding torch holder of the welding additive manufacturing head and is used to acquire vibration acceleration signals and / or angular velocity signals in three-dimensional space. The electrical parameter sampling unit is used to synchronously acquire welding current and / or arc voltage signals.

3. The containerized remote site emergency multi-functional repair cabin according to claim 1, characterized in that, The droplet transition mode recognition module includes a signal preprocessing unit and a mode classification unit. The signal preprocessing unit is connected to the multi-source signal acquisition module and is used to receive the vibration signal and electrical parameter signal acquired by the signal preprocessing unit, and to perform noise reduction and feature extraction on the vibration signal and electrical parameter signal. The mode classification unit is used to determine the current droplet transition mode type based on the extracted feature vector and a classification model.

4. The containerized remote site emergency multi-functional repair cabin according to claim 3, characterized in that, The signal preprocessing unit performs time-frequency domain transformation on the vibration signal to extract the energy distribution characteristics within a predetermined frequency band. The predetermined frequency band range includes at least one of the low-frequency impulse band, the mid-frequency continuous excitation band, and the high-frequency oscillation band.

5. The containerized remote site emergency multi-functional repair cabin according to claim 3, characterized in that, The classification model is a neural network-based classifier. Its input features include at least the frequency band energy features and electrical parameter waveform features of the vibration signal, and its output is the probability or category of short-circuit transition, droplet transition or jet transition.

6. The containerized remote site emergency multi-functional repair cabin according to claim 1, characterized in that, The adaptive adjustment module for process parameters includes an arc parameter adjustment unit connected to the arc additive manufacturing controller. The arc parameter adjustment unit is used to dynamically adjust the welding parameters according to the identified differences. The welding parameters include at least one of the following: peak pulse current, base current, pulse duty cycle, welding voltage, or wire feed speed.

7. The containerized remote site emergency multi-functional repair cabin according to claim 6, characterized in that, The process parameter adaptive adjustment module also includes a motion parameter adjustment unit, which is used to dynamically adjust the movement speed and / or path overlap rate of the robotic arm according to the intensity or frequency band characteristics of the vibration signal.

8. The containerized remote site emergency multi-functional repair cabin according to claim 1, characterized in that, The multi-source signal acquisition module also includes a welding wire surface condition detection unit, which is used to detect the oxidation degree of the welding wire surface online and send the detection result to the process parameter adaptive adjustment module; the process parameter adaptive adjustment module also includes a cleaning parameter adjustment unit connected to the arc additive manufacturing controller, which sends cleaning parameters to the arc additive manufacturing controller according to the detection result of the oxidation degree of the welding wire surface.

Images