The Design of Reflected Laser Intensity Testing System and Application of Quality Inspection for Laser Cladding Process

Abstract

Laser cladding is one of the critical technologies for additive manufacturing and rapid repair. Improving cladding performance by materials and process parameters is the leading research direction, but defects and instability of quality in the cladding process are inevitable. Therefore, it is necessary to study which factors are related to quality. In this paper, a new detection method is proposed to measure the radiation intensity of the reflected laser, laser scanning displacement, and temperature of the substrate while cladding. The characteristic values corresponding to the position of the cladding spots are extracted, the cladding quality is preliminarily evaluated and graded, and the correlation between them is verified with the method of machine learning nu-SVM. The results show that the accuracy of the model trained by 300 groups of data to predict the quality grades is 78.74%, which indicates that there is a strong correlation between these process variables and the cladding quality, and this method is feasible for the quality evaluation and control of the cladding process.

1. Introduction

Laser cladding, as an advanced additive manufacturing technology, has become a key technology in the manufacturing and repairing of complex parts due to the small heat-effected zone, low dilution rate, high machining accuracy, and strong operability in medical equipment, automobile manufacturing, and other fields [1,2,3]. Different from the traditional material-forming technology such as casting and forging, laser cladding uses a high-energy laser to melt the cladding material and the surface of the substrate to form a melting bath, and a metallurgically bonded cladding layer is formed after cooling [4,5,6]. Therefore, laser cladding technology is more used to strengthen metal materials’ surface to form an alloy layer with better properties on the substrate surface.

Properties of the surface after cladding are directly determined by the cladding materials, and many scholars prefer to study the improvement of properties by cladding different materials on the surface of a substrate. For instance, the microhardness and corrosion resistance of the AlSi5Cu1Mg alloy surface is remarkedly improved by laser cladding Ni-WC [7], and the AlNbTaZr_0.8 coatings of Ti6Al4V demonstrated the excellent resistance to high-temperature oxidation when compared with that of the substrate [8]. The process parameters of laser cladding (laser power, scanning speed, powder feeding rate/preset layer thickness, etc.) interact to directly affect the geometric morphology, microstructure, and mechanical properties of the cladding layer [9,10]. It can effectively improve performance and reduce defects by choosing appropriate process parameters, so scholars have conducted a lot of research on the optimization of laser cladding process parameters. Most studies focus on the influence of a certain parameter on the final cladding quality [11]; the orthogonal experiments are usually designed to determine the optimal combination of process parameters for some materials and parts [12]. For instance, the laser power has a more significant effect on the width and dilution ratio of the cladding layer, and the scanning speed has more on the height when cladding AISI stainless steel [13]. The microhardness and electrochemical corrosion resistance of the TC4 alloy cladding layer is also improved when the scanning speed increases to 10 mm/s [14]. The performance of the cladding layer is directly influenced by the cladding materials and process parameters, and these factors affect and restrict each other, so the properties and comprehensive quality are jointly determined.

Additionally, surface flatness, hardness, and mechanical strength, which are used to formulate the average quality and comprehensive performance, are affected by many other factors in the actual cladding process. The roughness and gloss of the substrate surface are mainly determined by the absorption of laser energy by the metal material [15]. Temperature causes the change of physical parameters of metal materials and affects the heat transfer and melting process, then affects the processing quality [16]. These variable parameters influence the decision-making mechanism of cladding process parameters on material properties and reflect the interaction between laser and materials. Accordingly, the differences in cladding quality are shown in different positions under the same process parameters as it is affected and reflected by the process variables.

Therefore, laser cladding is a complex process of multi-physical field coupling influenced comprehensively by multiple parameters. The relationship between process parameters, process variables, and cladding quality is studied to maintain the laser cladding quality as stable and consistent [17].

In this paper, three process variables (laser reflection intensity, laser scanning displacement, temperature) during cladding are proposed by analyzing the process of laser energy transfer and conversion, and a method for measuring radiation intensity of the reflected laser is invented to combine a laser cladding testing system in Section 2. The laser cladding experiment was carried out based on it. In Section 3, the correlation between three process variables and spot cladding quality is explored through the dispersion and quantification of pulse laser cladding quality and measured data. The conclusion is provided in Section 4.

2. Materials and Methods

2.1. Transfer and Conversion of Laser Energy

The laser energy is converted into three parts, absorbed, reflected, and transmitted, when it acts on any material surface. Its radiation energy must conform to the conservation of energy, which can be expressed below.

$$\alpha +\rho +\tau =1$$

(1)

where α is the absorption ratio, ρ is the reflection ratio, and τ is the transmission ratio, which represents the ratio of absorbed, reflected, and transmitted energy to the radiation energy of the incident laser.

For metal materials, the transmission ratio is 0 due to the lack of transparency. The energy of laser cladding always follows $\alpha +\rho =1$. Therefore, the effect of laser on materials can be explained from the two opposite sides of absorption and reflection.

The laser energy is absorbed partly by the surface of the material, which makes the internal lattice of metal vibrate and converts it into internal energy, and heats up to melt the substrate and cladding material to form a melting bath. Then, it is transferred to other parts of the substrate and the air through heat conduction and convection, which makes the temperature of substrate increase.

The mirror reflection and diffuse reflection of laser irradiation occur simultaneously on the uneven surface, and the input radiation is reflected with different intensity and energy density in all directions of the hemispherical space, which takes the laser reflection spot as the origin.

Therefore, according to the process of energy transfer and transformation, the energy conservation formula can be transformed into Formula (2).

$$Q_{input}=Q_{absorb}+Q_{reflect}={\displaystyle {\int }_{T_0}^{T}c(T)mdT+{\displaystyle \int {\displaystyle \underset{S}{\iint }I(\theta ,\phi )d\Omega }}}dt$$

(2)

where Q_input is the radiation energy of incident laser, c is the specific heat capacity (unit: J × kg⁻¹ × K⁻¹) of substrate that changes with temperature, T₀ is the initial temperature, T is the current temperature of the substrate, m is the mass of cladded sample, I(θ, φ) (unit: W/sr) is the distribution function of the radiation intensity of reflected laser along the zenith angle θ and distribution angle φ in the spherical coordinate system, which takes the laser reflection spot as the origin, dΩ is the differential solid angle in a certain direction (θ, φ), S is the upper hemisphere of the unit sphere, and t is the cladding time.

Temperature is a measure of the absorption of laser energy by the material according to Equation (2), which will determine the heat capacity and thermal conductivity of the substrate, thus affecting the cladding quality, so the temperature can reflect the cladding state.

The radiation intensity of the reflected laser in different directions of space is different for acting on the uneven surface, the laser reflection intensity reflects the material surface state and the surface quality after cladding. The laser will be reflected in all directions of hemispherical space with uniform radiation intensity when it irradiates the material surface with uniform roughness. However, the radiation intensity reflected in all directions of space is different for acting on the uneven preset layer. Therefore, the reflection intensity of the laser reflects the surface state of the substrate and the surface morphology after cladding.

The scanning displacement of laser cladding is also a factor affecting the cladding quality. Theoretically, stable scanning speed for the presetting laser cladding will result in the superposition of a series of cladding spots with equal center spacing. Still, the varying speed in the actual process will make the spot overlap rate higher or lower. A higher spot overlap rate means a larger overlap area, which leads to spatter of cladding materials and the formation of pores in the cladding area, and the lower overlap rate may lead to uncoated areas at the overlapping positions of spots and pores after reprocessing [18].

Therefore, it is feasible to evaluate or obtain feedback on the partial cladding quality through three groups of data: laser reflection intensity, laser scanning displacement, and substrate temperature. Based on it, a testing system for feedback data of laser cladding is designed.

2.2. Laser Cladding Test System

The laser cladding test system simultaneously measures laser reflection intensity, laser scanning displacement, and substrate temperature during the cladding process. The test system consists of three parts in Figure 1.

Radiation intensity sensors are the main part of the test system, which are used to measure the intensity of the laser reflected in the surrounding space when it acts on the substrate and cladding material. The bidirectional reflection distribution function (BRDF) [19] can be used to analyze the distribution of radiant energy of the reflected laser in all directions of space by the relationship between reflected light intensity and its zenith angle θ and distribution angle φ to study the absorption and reflection by metal materials in the actual process.

Mixtures of mirror reflection and diffuse reflection of laser irradiation are produced on the surface of the metal materials and spread to all parts of the space. The reflected laser directed to a small area of space can be regarded as a parallel beam. As shown in Figure 2, the reflected laser passes through the optical filter to filter out the space background light and part of the thermal radiation generated by the temperature rise of materials so that the beam reaching the condensing lens is closer to the reflected laser. Then the reflected laser passes through the lens and is focused on the receiving point of the avalanche photodiode (APD) located at the focal point of the lens.

A plurality of radiation intensity sensors is fixed everywhere in space, and the main structure is shown in Figure 3, in which the laser reflection spot, the center point (test point) of a condensing lens, and the receiving point of APD are located on the same line in the space.

Photodiodes convert the received optical signal into photocurrent, in addition, the current is proportional to the flux of received radiation and is closely related to the wavelength of radiation [20]. APD is the fastest, which amplifies the photocurrent by the avalanche multiplication effect of carriers. In Figure 3, the APD driving circuit amplifies the photocurrent into output voltage, which is collected by the acquisition card and saved in a computer.

Conventional sensors can be used to measure the substrate temperature and laser scanning displacement, and the collected data can also be recorded by the acquisition card or other data acquisition equipment.

2.3. Experiment

A 42CrMo plate with the size of 100 × 100 × 10 mm as substrate and T15 (high-speed steel) powder with an average particle size of 70 μm as cladding material were used in this laser cladding experiment. T15 powder was uniformly mixed with PVA solution (PVA as adhesive) and was laid on the surface of the substrate. In Figure 4, a preset layer with a size of 30 × 30 mm and a thickness of 0.18 mm was formed after being kept at 80 °C for 8 h.

The main equipment of the experiment is a pulsed fiber laser generator with a laser wavelength of 1064 nm, equipped with a three-axis ball screw linear guide to make the sample move on the horizontal plane and the laser head move vertically. This experiment is a single-layer cladding, where the laser generator is fixed, and the substrate moves at the speed v along the +Y axis, which can be regarded as laser scanning in the direction shown in Figure 2, because the spatial coordinates of the laser reflected spot on the substrate remain unchanged. The parameters in the laser cladding experiment are shown in

Table 1. Laser cladding parameters.

Peak Power P	Pulse Frequency f	Pulse Duration τ	Scanning Speed v	Track Spacing u
2.0 kW	10 Hz	4 ms	5 mm/s	0.455 mm

.

Characteristics of the radiation intensity sensor are shown in

Table 2. Characteristics of radiation intensity sensor.

Optical Filter (Bandpass Interference Filter)	Diameter d	25.4 mm
	Central Wave Length	1064 nm
	Full Width at Half Maximum	10 nm
Condensing Lens	Focal Length	50 mm
	Diameter d	25.4 mm
APD	Response Spectrum	400~1100 nm
	Response Time	0.5 ns

. The spherical coordinate system (r, θ, φ), which takes the laser reflected spot as the origin O, is established in Figure 5, and the coordinates of the six sensors are shown in

Table 3. Coordinates of radiation intensity sensors.

Test Point i	ri (cm)	θi (°)	φi (°)
1	24.0	64.321	310
2	34.2	28.343	0
3	46.5	25.700	250
4	69.5	60.617	200
5	41.6	44.443	160
6	62.5	56.605	130

, regarding the center of the condensing lens as the test point.

The laser displacement sensor is used to measure the pulsed laser scanning displacement, with a range of 70 mm, accuracy of 70 μm, and response time of 1.5 ms, and the output voltage is proportional to the measured displacement. Four K-shaped thermocouples (temperature sensors without an external power supply) are used to measure the temperature and are placed at the bottom of the substrate. The positions of these sensors are shown in Figure 6.

Output voltages of six radiation intensity sensors and one laser displacement sensor are collected by the acquisition card with dual chip and 16 channels and saved in the computer, and the card with the frequency of 20,000 Hz collects 7 groups of voltage data at the same time. The output voltages of four K-shaped thermocouples are collected by the data recorder with the frequency of 1000 Hz and directly recorded as the corresponding temperature value. The construction of the experimental site is shown in Figure 7.

3. Results and Discussion

3.1. Cladding Quality

The laser generator emits a pulse laser every 0.1 s, and the laser lasts for 4 ms. This laser irradiation causes a cladding point on the metal surface. Therefore, the defect-free result of single-layer cladding is the overlapping of a series of spots in Figure 8. The spot overlap rate depends on the track spacing u, scanning speed v, and spot diameter.

The cladding result with the peak power P of 2.0 kW is shown in Figure 9. There are some pores in the cladding layer. That is due to the poor drying effect, partial oxidation of powder materials, and residual water in the preset layer of PVA solution, resulting in gas being generated at a high temperature for laser irradiation and it is not discharged to form pores when the material melts and solidifies rapidly. Energy density at the overlap of cladding spots is increased in a short time, or the layer is unevenly paved and too thin in some positions [21], which makes it burned, broken down, splashed, and forms pores during cladding.

The cladding layer is discretized into a series of spots, and the cladding quality of spots is evaluated and divided into three grades:

H (high quality—complete cladding spot and no pore);

M (medium quality—broken cladding spot but no pore);

L (low quality—porous).

The example is shown in Figure 9. In this way, the preliminary quantification of the quality of each cladding spot is completed, which is convenient to analyze the correlation between the cladding quality and process variables.

3.2. Measured Data

A laser with a pulse frequency of 10 Hz irradiates the metal surface, so the radiation intensity sensors obtain the pulse signal every 0.1 s. In Figure 10, there is the radiant flux Φ (voltage value) of the reflected laser for 1.4 s measured at test point 2 during cladding. It can be seen that the intensity in the direction of the test point in the space changes due to the dynamic shift in laser absorptivity of the material and the difference in surface states in different positions of the preset layer.

Laser scanning displacement of the corresponding cladding position in the same period is shown in Figure 11. The linear guide rail displacement cannot completely coincide with its set displacement due to the stiffness of transmission parts and mechanical vibration, which also affects the spot overlap rate and cladding quality to a certain degree.

As above, the current state of the substrate and the absorption of laser energy is represented as temperature, and the temperature at different positions reflects the transmission of heat in the substrate.

3.3. Correlation Analysis

The comprehensive quality of the cladding layer is directly affected by the process parameters. Still, the quality in different positions strongly correlates with the dynamic process variables during cladding, such as laser energy absorption rate, spot overlap rate, temperature, etc. Therefore, the regional quality of the cladding layer is graded and quantified (quality of cladding spot), and the measured process variables (radiation intensity of reflected laser in different directions, laser scanning displacement, and temperature) are discretized to correspond with them one by one, to explore the correlation among them, or the regional cladding quality is evaluated through these process variables.

Discretize the measured data:

Quality grade Q of the cladding spot numbered k is obtained from the quality or defect of the cladding spot in Figure 9;

Radiation intensity L_i,k of reflected laser in different directions takes the peak value of its corresponding pulse signal as the characteristic value, that is the voltage value Φ_i,k (i is the mark of a test point in Figure 5) corresponding to the data in the black box as shown in Figure 10. The radiation intensity is calculated as Formula (3).

$${\mathrm{L}}_{i,k}=\frac{4{\Phi }_{i,k}{\mathrm{r}}_i{}^2}{\pi d^2}$$

(3)

where r_i is the distance from test point of radiation intensity sensor to laser reflection point in Figure 5 and Table 3; d is the diameter of condensing lens in Table 2.

Laser scanning displacement takes the error e_k between the measured displacement and the setting displacement as the characteristic value in Figure 11, and it is calculated as Formula (4).

$${\mathrm{e}}_{\mathrm{k}}=y-v(t-t_0)$$

(4)

where y is the output of displacement sensor; v is the scanning speed of laser in Table 1; t is the current time; and t₀ is the start time of this track.

The temperature at different positions of the substrate takes the temperature T_j,k (j is the mark of a temperature test point in Figure 6) at the current time as the characteristic value.

Therefore, the quality grades and 11 kinds of measured data of 427 cladding spots are obtained in the analysis of experimental results, and the data of 60 consecutive cladding points are shown in Figure 12.

A support vector machine (SVM) is a generalized linear classifier to classify data by supervised learning [22], and the data of 427 cladding spots (11 kinds of measured data as the input parameters and quality grade Q as the output label) are divided into a train set and a test set [23]. A total of 300 groups of data are trained by nu-SVM with radial basis function, and the labels of 127 groups are predicted through their input parameters. The classification of real tests and prediction are compared in Figure 13, and the accuracy rate is 78.74% (100/127). Therefore, there is a strong correlation between the process variables and cladding quality, and the method that the test system uses to respond to quality is feasible.

4. Conclusions

This paper proposes a test system for measuring the process variables (radiation intensity of reflected laser, laser scanning displacement, temperature of substrate) while cladding, and an experiment that cladded T15 powder on 42CrMo substrate was carried out in this system. To use Machine Learning to study the correlation between process variables and cladding quality, the measured data and cladding results were discretized and quantified. SVM was used to model 427 groups of data, 300 groups were trained, and the quality grades of 127 groups were predicted through the process variables. The training result indicates a strong correlation for the accuracy of 78.74% (100/127), and it is feasible to evaluate cladding quality through process variables.

For future work, more parameters can be studied, and more appropriate Machine Learning methods can be applied to obtain the exact relationship between cladding quality and process variables. Finally, the quality control of laser cladding can be completed by combining the process parameters and the state of the cladding process, and the cladding layer with stable quality and excellent performance can be formed.