High-precision temperature control method for laser

Abstract

The invention discloses a high-precision temperature control method for a laser, which comprises predictive temperature control and PID (Proportion Integration Differentiation) temperature control, and is characterized in that the predictive temperature control is used for controlling the flow rate of a refrigerant according to a preset temperature control model, and the preset temperature control model is established through machine learning by adopting a sample set obtained by historical working records of the laser; pID temperature control is PID control with the temperature of a backflow refrigerant as the feedback quantity and the flow speed of the refrigerant as the output quantity; when the difference between the backflow refrigerant temperature and the expected temperature of the temperature control model is smaller than a preset value, predictive temperature control is adopted; when the difference between the backflow refrigerant temperature and the expected temperature of the temperature control model is larger than a preset value, PID temperature control is adopted; the laser is cooled through the refrigerant, and the temperature of the refrigerant flowing into the laser is constant. The mode of constant refrigerant temperature, predictive temperature control and single PID refrigerant flow rate control is adopted, the refrigerating power of the water cooling machine and the laser power do not need to be coordinated, PID control parameters can be automatically adjusted, and high-precision control over the temperature of the laser is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of laser application, and in particular to a high-precision temperature control method for a laser. Background Art

[0002] Lasers are widely used in industrial production. Due to their own characteristics, lasers release a lot of heat when working. Generally, the operating temperature range of lasers is ±5~10℃. However, the heat (temperature) generated by lasers under actual working conditions far exceeds this range. Therefore, precise and constant temperature control is one of the technical obstacles to the application of lasers. Water chillers are a widely used cooling mechanism and are also widely used for constant temperature control of lasers. Water chillers generally use cooling water / liquid as a refrigerant. The low-temperature water output by the water chiller flows through the internal circulation pipeline of the laser and takes away the heat before returning to the water chiller. The water chiller monitors the return water temperature and adjusts the cooling capacity to stabilize the high-temperature return water temperature near the set temperature. Existing constant temperature control for lasers usually adopts a dual PID control method, which has the following defects:

[0003] Because temperature control is a typical time-delay control system, changes in laser power immediately cause changes in heat generation. However, it takes time for this heat change to be detected by the water chiller through the refrigerant temperature rise. By the time the water chiller adjusts its cooling power based on the detected temperature change, the laser's output power has already changed several times. This causes the water chiller's cooling power to become uncoordinated with the laser power, causing PID (Proportional-Integral-Derivative) parameter oscillation. Summary of the Invention

[0004] To overcome the above-mentioned defects, the present invention aims to provide a high-precision temperature control method for a laser, which adopts a constant refrigerant temperature, predictive temperature control and single PID control of the refrigerant flow rate. There is no need to coordinate the cooling power of the water chiller with the laser power, and the PID control parameters can be automatically adjusted, thereby achieving high-precision control of the laser temperature.

[0005] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0006] A high-precision temperature control method for a laser, including predictive temperature control and PID temperature control. The predictive temperature control controls the refrigerant flow rate according to a preset temperature control model. The preset temperature control model is established through machine learning using a sample set obtained from the laser's historical operating records.

[0007] The PID temperature control is a PID control with the return refrigerant temperature as the feedback quantity and the refrigerant flow rate as the output quantity;

[0008] When the difference between the return refrigerant temperature and the expected temperature of the temperature control model is less than a predetermined value, predictive temperature control is adopted; when the difference between the return refrigerant temperature and the expected temperature of the temperature control model is greater than a predetermined value, PID temperature control is adopted;

[0009] The laser is cooled by the refrigerant, and the temperature of the refrigerant flowing into the laser is constant.

[0010] Preferably, the PID temperature control comprises the following steps:

[0011] a. Obtain the return refrigerant temperature and sample the return refrigerant temperature through a temperature sensor;

[0012] b. Cold night flow control: according to the return refrigerant temperature, a PID controller is used to control the refrigerant flow rate in real time so that the return refrigerant temperature is maintained within a preset temperature range;

[0013] c. Loop through steps a and b;

[0014] In step b, the sampling frequency of the temperature sensor is T s , the refrigerant flow rate measured at each sampling time k is F actual [k], the target refrigerant flow rate is recorded as F target , calculate the error e[k]:

[0015] e[k]=F target -F actual [k] (1)

[0016] A closed-loop control algorithm based on PID control is used to generate a control instruction u[k] to adjust the control parameters of the refrigerant flow rate in real time, thereby compensating for the error:

[0017]

[0018] Among them, K p is the proportional gain, K i is the integral gain, K d is the differential gain;

[0019] The PID controller generates a control signal u[k], which is used to adjust the control parameters of the refrigerant so that the refrigerant flow rate gradually approaches the target value:

[0020]

[0021] Furthermore, there are a plurality of temperature sensors, which are arranged at equal intervals along the refrigerant return path. Step a further includes processing the temperature sensor signals as follows:

[0022] Assume the true signal of the temperature sensor is S j(t), the measurement noise is N j (t), the total number of temperature sensors is N, then the continuous output signal of the temperature sensor can be described as:

[0023] X j (t) = S j (t)+N j (t),j=1,2,...,N. (4)

[0024] Unify the sampling period of all sensors to T s , k is used to represent the sampling point index during discrete sampling, that is, the kth sampling point, then the discrete sampling time t k Defined as:

[0025] t k =k·T s , k=0,1,2,.... (5)

[0026] At each sampling moment, data is collected from each temperature sensor to obtain discrete sampling data:

[0027] X j [k] = X j (t k )=S j (t k )+N j (t k ) (6)

[0028] The return refrigerant temperature is the arithmetic average of the temperatures measured by all temperature sensors.

[0029] Furthermore, step a also includes filtering the temperature sensor signal to eliminate high-frequency noise.

[0030] Preferably, the filtering method is wavelet transform.

[0031] Preferably, the temperature control model is a neural network model, the input of the temperature control model includes the return refrigerant temperature and the temperature rise rate, and the output of the temperature control model is the refrigerant flow rate.

[0032] Preferably, the temperature rise rate is the rate of change of the return refrigerant temperature between adjacent sampling moments, and the calculation method is as follows:

[0033]

[0034] Among them, s t is the change rate of the return refrigerant temperature at sampling time t, v t is the return refrigerant temperature at sampling time t, v t-1is the return refrigerant temperature at sampling time t-1, and T is the sampling interval.

[0035] The beneficial effects of the present invention are as follows:

[0036] 1. The present invention adopts a constant refrigerant temperature, predictive temperature control and single PID control of the refrigerant flow rate. There is no need to coordinate the cooling power of the water chiller with the laser power, and it can automatically adjust the PID control parameters to achieve high-precision control of the laser temperature.

[0037] 2. The present invention uses multiple temperature sensors, which can effectively eliminate the error of a single temperature sensor and the influence of environmental factors. DETAILED DESCRIPTION

[0038] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is described in further detail.

[0039] Example 1

[0040] This embodiment discloses a high-precision temperature control method for a laser, including predictive temperature control and PID temperature control. The predictive temperature control controls the refrigerant flow rate according to a preset temperature control model. The preset temperature control model is established through machine learning using a sample set obtained from the laser's historical operating records.

[0041] PID temperature control is a PID control that uses the return refrigerant temperature as feedback and the refrigerant flow rate as output;

[0042] When the difference between the return refrigerant temperature and the expected temperature of the temperature control model is less than a predetermined value, predictive temperature control is adopted; when the difference between the return refrigerant temperature and the expected temperature of the temperature control model is greater than a predetermined value, PID temperature control is adopted;

[0043] The laser is cooled by the refrigerant, and the temperature of the refrigerant flowing into the laser is constant.

[0044] Among them, PID temperature control includes the following steps:

[0045] a. Obtain the return refrigerant temperature and sample the return refrigerant temperature through a temperature sensor;

[0046] b. Cold night flow control: according to the return refrigerant temperature, a PID controller is used to control the refrigerant flow rate in real time so that the return refrigerant temperature is maintained within a preset temperature range;

[0047] c. Loop through steps a and b;

[0048] In step b, the sampling frequency of the temperature sensor is T s , the refrigerant flow rate measured at each sampling time k is F actual [k], the target refrigerant flow rate is recorded as Ftarget , calculate the error e[k]:

[0049] e[k]=F target -F actual [k] (1)

[0050] In order to cope with the nonlinearity and uncertainty of the on-site environment, an adaptive correction algorithm is introduced to compare the real-time measured refrigerant flow rate with the target preload force. A closed-loop control algorithm based on PID control is used to generate a control instruction u[k] to adjust the control parameters of the refrigerant flow rate in real time, thereby compensating for the error:

[0051]

[0052] Among them, K p is the proportional gain, K i is the integral gain, K d is the differential gain;

[0053] The PID controller generates a control signal u[k], which is used to adjust the control parameters of the refrigerant so that the refrigerant flow rate gradually approaches the target value:

[0054]

[0055] The temperature control model is a neural network model. The input of the temperature control model includes the return refrigerant temperature and the temperature rise rate. The output of the temperature control model is the refrigerant flow rate. The temperature rise rate is the rate of change of the return refrigerant temperature between adjacent sampling times. It is calculated as follows:

[0056]

[0057] Among them, s t is the change rate of the return refrigerant temperature at sampling time t, v t is the return refrigerant temperature at sampling time t, v t-1 is the return refrigerant temperature at sampling time t-1, and T is the sampling interval.

[0058] Example 2

[0059] Based on Example 1, considering the error of the temperature sensor and the influence of the environment, this embodiment adopts a plurality of temperature sensors, which are arranged at equal intervals along the refrigerant return path. Step a also includes processing the temperature sensor signal as follows:

[0060] Assume the true signal of the temperature sensor is S j (t), the measurement noise is N j (t), the total number of temperature sensors is N, then the continuous output signal of the temperature sensor can be described as:

[0061] X j (t) = S j (t)+N j (t),j=1,2,...,N. (4)

[0062] Unify the sampling period of all sensors to T s , k is used to represent the sampling point index during discrete sampling, that is, the kth sampling point, then the discrete sampling time t k Defined as:

[0063] t k =k·T s , k=0,1,2,.... (5)

[0064] At each sampling moment, data is collected from each temperature sensor to obtain discrete sampling data:

[0065] X j [k] = X j (t k )=S j (t k )+N j (t k ) (6)

[0066] The return refrigerant temperature is the arithmetic average of the temperatures measured by all temperature sensors.

[0067] Step a also includes filtering the temperature sensor signal using a wavelet transform to eliminate high-frequency noise and improve signal quality.

[0068] The rest of this embodiment is the same as that of Embodiment 1, and therefore will not be described in detail.

[0069] Of course, the present invention may have many other embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art may make various corresponding changes and modifications based on the present invention, but these corresponding changes and modifications should all fall within the scope of protection of the claims attached to the present invention.

Claims

1. A high-precision temperature control method for a laser, characterized in that: It includes predictive temperature control and PID temperature control. The predictive temperature control controls the refrigerant flow rate according to a preset temperature control model. The preset temperature control model is established through machine learning using a sample set obtained from the historical working records of the laser. The PID temperature control is a PID control with the return refrigerant temperature as the feedback quantity and the refrigerant flow rate as the output quantity; When the difference between the return refrigerant temperature and the expected temperature of the temperature control model is less than a predetermined value, predictive temperature control is adopted; when the difference between the return refrigerant temperature and the expected temperature of the temperature control model is greater than a predetermined value, PID temperature control is adopted; The laser is cooled by the refrigerant, and the temperature of the refrigerant flowing into the laser is constant.

2. The high-precision temperature control method for a laser according to claim 1, characterized in that: The PID temperature control comprises the following steps: a. Obtain the return refrigerant temperature and sample the return refrigerant temperature through a temperature sensor; b. Cold night flow control: according to the return refrigerant temperature, a PID controller is used to control the refrigerant flow rate in real time so that the return refrigerant temperature is maintained within a preset temperature range; c. Loop through steps a and b; In step b, the sampling frequency of the temperature sensor is T s , the refrigerant flow rate measured at each sampling time k is F actual [k], the target refrigerant flow rate is recorded as F target , calculate the error e[k]: e[k]=F target -F actual [k] (1) A closed-loop control algorithm based on PID control is used to generate a control instruction u[k] to adjust the control parameters of the refrigerant flow rate in real time, thereby compensating for the error: Among them, K p is the proportional gain, K i is the integral gain, K d is the differential gain; The PID controller generates a control signal u[k], which is used to adjust the control parameters of the refrigerant so that the refrigerant flow rate gradually approaches the target value:

3. The high-precision temperature control method for a laser according to claim 2, characterized in that: There are a plurality of temperature sensors, which are arranged at equal intervals along the refrigerant return path. Step a further includes processing the temperature sensor signals as follows: Assume the true signal of the temperature sensor is S j (t), the measurement noise is N j (t), the total number of temperature sensors is N, then the continuous output signal of the temperature sensor can be described as: X j (t)=S j (t)+N j (t),j=1,2,...,N. (4) Unify the sampling period of all sensors to T s , k is used to represent the sampling point index during discrete sampling, that is, the kth sampling point, then the discrete sampling time t k Defined as: t k =k·T s ,k=0,1,2,.... (5) At each sampling moment, data is collected from each temperature sensor to obtain discrete sampling data: X j [k]=X j (t k )=S j (t k )+N j (t k ) (6) The return refrigerant temperature is the arithmetic average of the temperatures measured by all temperature sensors.

4. The high-precision temperature control method for a laser according to claim 3, characterized in that: Step a also includes filtering the temperature sensor signal to eliminate high-frequency noise.

5. The high-precision temperature control method for a laser according to claim 4, characterized in that: The filtering method is wavelet transform.

6. The high-precision temperature control method for a laser according to any one of claims 1 to 5, characterized in that: The temperature control model is a neural network model. The input of the temperature control model includes the return refrigerant temperature and the temperature rise rate, and the output of the temperature control model is the refrigerant flow rate.

7. The high-precision temperature control method for a laser according to claim 4, characterized in that: The temperature rise rate is the rate of change of the return refrigerant temperature between adjacent sampling moments, and its calculation method is as follows: Among them, s t is the change rate of the return refrigerant temperature at sampling time t, v t is the return refrigerant temperature at sampling time t, v t-1 is the return refrigerant temperature at sampling time t-1, and T is the sampling interval.

Images