A power control method, system and device for semiconductor laser

By analyzing the similarity of current value distribution of semiconductor lasers, screening similar sequences, determining data correlation intensity, and adaptively adjusting the smoothing coefficient, the problem of current instability in semiconductor laser power control is solved, and prediction accuracy and control effect are improved.

CN118589294BActive Publication Date: 2024-11-15DALIAN ZHONGKE SUPER SILICON INTEGRATION TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411068013.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-06
Publication Date
2024-11-15
Estimated Expiration
2044-08-06

AI Technical Summary

Technical Problem

The prior art in semiconductor laser power control is instable due to temperature changes, and the prediction accuracy of the exponential moving average algorithm with fixed smoothing coefficient is low, so it is impossible to effectively control the power of the semiconductor laser in a timely and effective manner.

Method used

Through the similarity analysis between the current value prediction sequence and historical data, the current value similarity sequence is screened out, the data correlation intensity is determined, and the smoothing coefficients in the exponential moving average algorithm are adaptively adjusted to achieve more accurate current value prediction and power control.

Benefits of technology

The accuracy and stability of semiconductor laser power control are improved, and more accurate real-time current prediction data values ​​are obtained through adaptive smoothing coefficient combined with exponential moving average algorithm, thereby achieving efficient control of semiconductor laser power.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118589294B_ABST
    Figure CN118589294B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of laser control technology, and in particular to a power control method, system and device for a semiconductor laser. The present invention selects a current value similarity sequence by the similarity of current value distribution between a current value prediction sequence and each current value reference sequence; thereby, according to the correlation between each current value similarity sequence and the reference similarity, the data correlation strength of each current data corresponding to the current value prediction sequence is determined, thereby obtaining a more accurate adaptive smoothing coefficient corresponding to an exponential moving average algorithm based on the data correlation strength, so that when the current value is predicted by combining the adaptive smoothing coefficient with the exponential moving average algorithm, the real-time current prediction data value obtained is more accurate; correspondingly, the accuracy of semiconductor laser power control performed based on the predicted real-time current prediction data value is higher.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of laser control technology, and in particular to a power control method, system and equipment for a semiconductor laser. Background Art

[0002] Semiconductor lasers, or laser diodes, are devices that use semiconductor materials as working materials to generate lasers, and emit lasers through electrical excitation methods such as current injection. For semiconductor lasers, precise power control can keep the semiconductor lasers in the best working state, making the electro-optical conversion efficiency, working performance and stability of semiconductor lasers higher.

[0003] Since the semiconductor laser is the light source of the interference system, it has energy-consuming properties, that is, it will continuously release heat during the operation of the semiconductor laser, causing the temperature of the semiconductor laser to rise, thereby causing the resistance inside the semiconductor laser to change, and the corresponding current value will also change accordingly, thereby indirectly affecting the power of the semiconductor laser; therefore, the power change of the semiconductor laser will be unstable due to the influence of temperature. Therefore, in order to ensure the stable state of the power of the semiconductor laser, it is necessary to monitor the current or voltage of the semiconductor laser, and control the power of the semiconductor laser in real time by adjusting the voltage or current.

[0004] Considering that the power control of semiconductor lasers based on real-time voltage or current has a certain delay, it is impossible to timely control the power of semiconductor lasers in real time. Therefore, the prior art usually uses an exponential moving average algorithm to predict data based on historical current data, so as to control the power of semiconductor lasers more timely based on the predicted current prediction data value; however, due to the release of heat energy in the working state of the semiconductor, the resistance of the semiconductor changes, causing the current to fluctuate, and the corresponding semiconductor laser current has different numerical distributions in the local time period, resulting in the use of a fixed smoothing coefficient exponential moving average algorithm for prediction, which will make the predicted real-time current prediction data value not accurate enough, making the accuracy of semiconductor laser power control based on the predicted real-time current prediction data value low. Summary of the invention

[0005] The present application provides a power control method, system, and device for a semiconductor laser. The method screens out a current value similarity sequence by analyzing the similarity of current value distribution between a current value prediction sequence and each current value reference sequence. The method determines the data correlation strength of each current data corresponding to the current value prediction sequence according to the correlation between each current value similarity sequence and the reference similarity. The method obtains a more accurate adaptive smoothing coefficient corresponding to an exponential moving average algorithm based on the data correlation strength. The method solves the problem that when an exponential moving average algorithm with a fixed smoothing coefficient is used for prediction, the predicted real-time current prediction data value is not accurate enough. The method makes the real-time current prediction data value more accurate when the current value is predicted by combining an adaptive smoothing coefficient with an exponential moving average algorithm. The method makes the power control of the semiconductor laser based on the predicted real-time current prediction data value more accurate.

[0006] The first aspect of the present application provides a power control method for a semiconductor laser, comprising:

[0007] Collecting a current value prediction sequence representing neighborhood data at the current moment and a current value base sequence representing historical data of the semiconductor laser in a working state; traversing the current value base sequence through a sliding window with the same length as the current value prediction sequence to obtain at least two current value reference sequences;

[0008] Determine the reference similarity between each current value reference sequence and the current value prediction sequence in the dimension of the current value change trend; filter out the current value similarity sequences according to the reference similarity; determine the data association strength of the current value prediction sequence according to the association between the reference similarity corresponding to each current value similarity sequence and each current value similarity sequence;

[0009] The priori smoothing coefficient in the exponential moving average algorithm is weighted by the data association strength to determine the adaptive smoothing coefficient of the current value prediction sequence; the current value is predicted by combining the adaptive smoothing coefficient with the exponential moving average algorithm to determine the real-time current prediction data value;

[0010] The semiconductor laser power is controlled according to the real-time current prediction data value.

[0011] In an embodiment of the first aspect of the present application, the method for obtaining the reference similarity includes:

[0012] Reversely arrange the current values ​​in each current value reference sequence to obtain a corresponding current value reverse sequence;

[0013] taking each current value reference sequence and each current value reverse sequence as a target sequence in turn;

[0014] The difference between the current value at each index value in the target sequence and the current value at the same index value in the current value prediction sequence is used as a reference predicted current difference corresponding to each index value in the target sequence;

[0015] The difference between the mean of the reference predicted current differences corresponding to all index values ​​in the target sequence and the reference predicted current difference corresponding to each index value is used as the predicted difference standard deviation corresponding to each index value in the target sequence;

[0016] Performing negative correlation normalization on the accumulated values ​​of the prediction difference standard deviations corresponding to all index values ​​in the target sequence to determine the initial similarity between the target sequence and the current value prediction sequence;

[0017] The maximum value of the initial similarity between each current value reference sequence and the corresponding current value reverse sequence is used as the reference similarity between each current value reference sequence and the current value prediction sequence.

[0018] In an embodiment of the first aspect of the present application, the method for acquiring a similar sequence of current values ​​includes:

[0019] The current value reference sequence whose corresponding reference similarity is greater than a preset similarity threshold is used as the current value screening sequence;

[0020] When the initial similarity of the current value screening sequence is less than the initial similarity of the corresponding current value reverse sequence, taking the current value reverse sequence corresponding to the current value screening sequence as the current value similarity sequence;

[0021] When the initial similarity of the current value screening sequence is greater than or equal to the initial similarity of the corresponding current value reverse sequence, the current value screening sequence is used as a current value similarity sequence.

[0022] In an embodiment of the first aspect of the present application, the method for acquiring data association strength includes:

[0023] Arrange the reference similarities of all current value similarity sequences in the order in which the corresponding current value reference sequences are obtained to obtain a reference similarity sequence;

[0024] Taking each current value similarity sequence as a row of the matrix, a current value similarity matrix is ​​constructed according to all current value similarity sequences;

[0025] Obtaining a column vector composed of elements in each column of the current value similarity matrix; using the cosine similarity between each column vector and the reference similarity sequence as the local correlation degree of each column vector;

[0026] The normalized value of the mean of the local correlation degree of all column vectors is used as the data correlation strength of the current value prediction sequence.

[0027] In an embodiment of the first aspect of the present application, the method for obtaining the adaptive smoothing coefficient includes:

[0028] Determining a corresponding adaptive smoothing coefficient according to the product of the data association strength and the priori smoothing coefficient;

[0029] In an embodiment of the first aspect of the present application, the method for obtaining the real-time current prediction data value includes:

[0030] After the smoothing coefficient in the exponential moving average algorithm is replaced by the adaptive smoothing coefficient, the current value prediction sequence is used as the input of the exponential moving average algorithm to output the real-time current prediction data value at the current moment.

[0031] In an embodiment of the first aspect of the present application, the method for controlling semiconductor laser power according to the real-time current prediction data value includes:

[0032] The difference between the real-time current prediction data value and the current value at the current moment is used as the current signal prediction deviation;

[0033] Using the ratio between the current signal prediction deviation and the real-time current prediction data value as a current signal fluctuation reference value;

[0034] When the current signal fluctuation reference value is greater than a preset fluctuation threshold, the power of the semiconductor laser is controlled by adjusting the current value of the semiconductor laser to the real-time current prediction data value;

[0035] When the current signal fluctuation reference value is less than or equal to the preset fluctuation threshold value, the current value of the semiconductor laser is not adjusted.

[0036] In an embodiment of the first aspect of the present application, the method of adjusting the current value of the semiconductor laser to the real-time current prediction data value includes:

[0037] The real-time current prediction data value is fed back to the power control module of the semiconductor laser, so that the power control module adjusts the current value of the semiconductor laser to the real-time current prediction data value through the pulse width modulation technology.

[0038] In a second aspect, the present application provides a power control system for a semiconductor laser, the system comprising:

[0039] A data acquisition preprocessing module is used to acquire a current value prediction sequence representing neighborhood data at the current moment and a current value base sequence representing historical data of the semiconductor laser in a working state; traverse the current value base sequence through a sliding window with the same length as the current value prediction sequence to obtain at least two current value reference sequences;

[0040] A first determination module is used to determine the reference similarity between each current value reference sequence and the current value sequence in the dimension of the current value change trend; filter out the current value similarity sequences in all the current value reference sequences according to the reference similarity; and determine the data association strength of the current value prediction sequence according to the association between the reference similarity corresponding to each current value similarity sequence and each current value similarity sequence;

[0041] The second determination module is used to weight the priori smoothing coefficient in the exponential moving average algorithm with the data association strength to determine the adaptive smoothing coefficient of the current value prediction sequence; and determine the real-time current prediction data value of the current value prediction sequence by combining the adaptive smoothing coefficient with the exponential moving average algorithm;

[0042] The semiconductor laser power control module is used to perform semiconductor laser power control according to the real-time current prediction data value.

[0043] In a third aspect, the present application provides a computer device comprising a memory and a processor; the memory is used to store computer program code that can be executed on the processor, and the processor is used to call and execute the computer program code from the memory to execute a method as described in the first aspect of the present application or any embodiment of the first aspect.

[0044] In a fourth aspect, the present application provides a computer program product, comprising a computer program code, which, when executed, performs the method of the first aspect of the present application or any embodiment of the first aspect.

[0045] In a fifth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program code. When the computer program code is executed, the method of the first aspect or any embodiment of the first aspect of the present application is executed.

[0046] This application has the following beneficial effects:

[0047] By analyzing the similarity between the current value reference sequence and the current value prediction sequence on the current value distribution corresponding to the semiconductor laser, the current value similar sequence is screened out, so as to accurately evaluate the overlap between the historical current data and the predicted current data according to the correlation between the reference similarity corresponding to each current value similar sequence and each current value similar sequence, so as to quantify the correlation between the current value data, so that the obtained data correlation strength can be combined with the numerical distribution of the semiconductor laser current in different local time periods, and the corresponding data correlation strength is more accurate, that is, the adaptive smoothing coefficient measured by the data correlation strength of the current value prediction sequence is more accurate, so that the real-time current prediction data value predicted by the exponential moving average algorithm combined with the current value prediction sequence is more accurate, so that the accuracy of semiconductor laser power control based on the predicted real-time current prediction data value is higher. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0049] Figure 1 A flow chart of a power control method for a semiconductor laser provided by an embodiment of the present invention;

[0050] Figure 2 A structural diagram of a power control system of a semiconductor laser provided by one embodiment of the present invention;

[0051] Figure 3 A schematic diagram of the structure of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the power control method, system, and device of a semiconductor laser proposed in accordance with the present invention, as well as its specific implementation method, structure, characteristics, and effects are described in detail below in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment, and the specific features, structures, or characteristics in one or more embodiments may be combined in any suitable form. In addition, the terms "first" and "second" are used only for descriptive purposes and cannot be understood as implying or suggesting relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features.

[0053] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0054] The specific scheme of a power control method, system and device for a semiconductor laser provided by the present invention is described in detail below with reference to the accompanying drawings.

[0055] The present invention provides a method for controlling the power of a semiconductor laser. Figure 1 , which shows a flow chart of a power control method for a semiconductor laser provided by an embodiment of the present invention, the method comprising:

[0056] Step S101: Collect a current value prediction sequence representing neighborhood data at the current moment and a current value base sequence representing historical data of the semiconductor laser in a working state; traverse the current value base sequence through a sliding window with the same length as the current value prediction sequence to obtain at least two current value reference sequences.

[0057] In a specific implementation of the embodiment of the present invention, an ammeter is installed on the semiconductor laser, and in the working state of the semiconductor laser, a current value is collected at every sampling moment, the sampling moment interval is set to 1 second, and the total collection time is set to 5 minutes; thereby obtaining the current value history data of the semiconductor laser at the current moment and before the current moment in the working state; after arranging all the current value history data in chronological order, a current value history sequence is obtained; the last preset number of data in the current value history sequence is used as the current value prediction sequence; and the other current value history data outside the current value prediction sequence is used as the current value base sequence. It should be noted that the implementer can also collect current value data through other devices according to the specific implementation environment, such as current sensors; and the implementer can adjust the sampling moment interval and the total collection time according to the specific implementation environment; and in a specific implementation of the embodiment of the present invention, the preset number is set to 10, and the implementer can adjust it according to the specific implementation environment.

[0058] The current value prediction sequence is a sequence used for exponential moving average algorithm prediction. For the current value prediction sequence, the more similar it is to the local data in the current value base sequence, the more the current data of the semiconductor laser has certain repeatability and periodicity in time series, that is, the higher the correlation degree of the current data, which affects the determination of the subsequent adaptive smoothing coefficient. Therefore, it is necessary to perform similarity analysis on the current value prediction sequence and each local data in the current value base sequence; therefore, it is first necessary to obtain the current value sequence corresponding to each local data. In order to make the obtained local data more comprehensive, the current value base sequence is traversed through the sliding window, so that all the current value reference sequences obtained can more comprehensively represent the local data, so that the subsequent similarity analysis is more complete. It should be noted that the sliding window traversal is a prior art well known to those skilled in the art, and will not be further described here; and the purpose of setting the sliding window length to be the same as the current value prediction sequence length is to make the obtained current value reference sequence and the current value prediction sequence length the same, so as to facilitate the similarity measurement.

[0059] Step S102: Determine the reference similarity between each current value reference sequence and the current value prediction sequence in the dimension of the current value change trend; filter out the current value similarity sequences according to the reference similarity; determine the data association strength of the current value prediction sequence according to the association between the reference similarity corresponding to each current value similarity sequence and each current value similarity sequence.

[0060] After obtaining each current value reference sequence through step S101, further according to the purpose of the embodiment of the present invention, the reference similarity between each current value reference sequence and the current value prediction sequence is determined in the dimension of the current value change trend, so that the reference similarity can be combined with the local current value distribution, so that the data association strength obtained subsequently can include the local current value distribution characteristics, thereby combining the different characteristics of the numerical distribution of the semiconductor laser current in the local time period, so that the real-time current prediction data value obtained subsequently is more accurate. After the current value similarity sequence is screened out according to the reference similarity, the data association strength corresponding to the current value prediction sequence, which represents the association with the historical current value data, is determined based on the association between the reference similarity corresponding to each current value similarity sequence and each current value similarity sequence, reflecting the ability of the current value prediction sequence to capture the historical current value change trend, thereby determining a more accurate adaptive smoothing coefficient.

[0061] Preferably, in some possible implementations of the embodiments of the present invention, the method for obtaining the reference similarity in step S102 includes:

[0062] The current values ​​in each current value reference sequence are arranged in reverse order to obtain the corresponding current value reverse sequence; each current value reference sequence and each current value reverse sequence are taken as the target sequence in turn; the difference between the current value under each index value in the target sequence and the current value under the same index value in the current value prediction sequence is taken as the reference predicted current difference corresponding to each index value in the target sequence; the difference between the mean of the reference predicted current differences corresponding to all index values ​​in the target sequence and the reference predicted current difference corresponding to each index value is taken as the standard deviation of the predicted difference corresponding to each index value in the target sequence; the accumulated value of the predicted difference standard deviation corresponding to all index values ​​in the target sequence is negatively normalized to determine the initial similarity between the target sequence and the current value prediction sequence; the maximum value of the initial similarity between each current value reference sequence and the corresponding current value reverse sequence is taken as the reference similarity between each current value reference sequence and the current value prediction sequence.

[0063] Because the current signal data may have anti-coincidence phenomenon, in order to take into account the occurrence of anti-coincidence phenomenon, when calculating the initial similarity of each current value reference sequence, the initial similarity of the corresponding current value reverse sequence is calculated at the same time, and the maximum value of the initial similarity between the current value reference sequence and the corresponding current value reverse sequence is selected as the reference similarity; and in subsequent analysis, the sequence with the larger initial similarity between the current value reference sequence and the corresponding current value reverse sequence is also used as the current value similarity sequence for analysis. For example, when the current value reference sequence is When , the corresponding current value reverse sequence is .

[0064] For the calculation process of the initial similarity, its basic principle is to calculate the difference in current values ​​under the same index value in the target sequence and the current value prediction sequence, which is similar to the DTW distance calculation process in the dynamic time warping algorithm. The larger the corresponding reference predicted current difference, the larger the overall numerical difference between the target sequence and the current value prediction sequence. Considering that the overall difference is consistent in the region, it can also reflect the same change trend. Therefore, the cumulative value of the mean deviation of all reference predicted current differences is further calculated to reflect the distribution trend difference between the target sequence and the current value prediction sequence; considering that the larger the distribution trend difference, the lower the corresponding similarity, negative correlation normalization is further performed to make the obtained initial similarity more accurate. In other possible implementation methods of the embodiments of the present invention, the variance of the reference predicted current differences corresponding to all index values ​​in the target sequence can be used to replace the cumulative value of the predicted difference standard deviation to determine the corresponding initial similarity, which all characterize the fluctuation chaos characteristics and will not be further elaborated here.

[0065] In a specific implementation of the embodiment of the present invention, the process of obtaining the initial similarity is expressed by the formula: ;in, The target sequence The initial similarity with the current value prediction sequence; The target sequence The number of index values ​​in , that is, the length of the target sequence and the current value prediction sequence; The target sequence Middle The current value under the index value and the current value prediction sequence The difference between the current values ​​under the index value, that is, the target sequence Middle The reference predicted current difference corresponding to the index value; is the mean of the reference prediction current differences corresponding to all index values ​​in the target sequence; The target sequence Middle The standard deviation of the predicted differences corresponding to the index values; is the absolute value symbol; To normalize the function, the normalization method uses linear normalization.

[0066] Preferably, in some possible implementations of the embodiments of the present invention, the method for acquiring the current value similarity sequence in step S102 includes:

[0067] A current value reference sequence whose corresponding reference similarity is greater than a preset similarity threshold is used as a current value screening sequence; when the initial similarity of the current value screening sequence is less than the initial similarity of the corresponding current value reverse sequence, the current value reverse sequence corresponding to the current value screening sequence is used as a current value similarity sequence; when the initial similarity of the current value screening sequence is greater than or equal to the initial similarity of the corresponding current value reverse sequence, the current value screening sequence is used as a current value similarity sequence.

[0068] Due to the influence of the anti-overlapping phenomenon of current signal data, it is necessary to calculate the initial similarity of each current value reference sequence and the initial similarity of its corresponding inverted current value reverse sequence, among which the largest initial similarity can represent the reference similarity between the current value change trend of the corresponding data segment and the predicted current value sequence. Therefore, it is necessary to further judge whether the reference similarity comes from the current value reference sequence or the current value reverse sequence, so as to obtain a current similarity sequence with a more similar corresponding current change trend.

[0069] In another specific implementation of the embodiment of the present invention, a k-means clustering analysis is performed on the reference similarities corresponding to all current value reference sequences, and the number of clusters, i.e., the k value, is set to 2, to obtain two reference similarity clusters; the reference similarity mean of each reference similarity cluster is calculated, and the current value reference sequences corresponding to all reference similarities in the reference similarity cluster with the largest reference similarity mean are used as current value screening sequences. It should be noted that the k-means clustering analysis algorithm is an existing technology well known to those skilled in the art, and the implementer can adjust the number of clusters according to the specific implementation environment, which will not be further elaborated here.

[0070] Preferably, in some possible implementations of the embodiments of the present invention, the method for acquiring the data association strength in step S102 includes:

[0071] The reference similarities of all current value similarity sequences are arranged in the order of obtaining the corresponding current value reference sequences to obtain a reference similarity sequence; each current value similarity sequence is taken as a row of the matrix, and a current value similarity matrix is ​​constructed according to all current value similarity sequences; a column vector composed of elements in each column of the current value similarity matrix is ​​obtained; the cosine similarity between each column vector and the reference similarity sequence is used as the local correlation degree of each column vector; and the normalized value of the mean of the local correlation degrees of all column vectors is used as the data correlation strength of the current value prediction sequence.

[0072] The higher the similarity between the column vector in the current value similarity matrix and the reference similarity sequence, the higher the correlation between the elements of the column vector, and the greater the corresponding local correlation. Therefore, the local correlation degrees of all column vectors are fused so that the obtained data correlation strength can characterize the data correlation degree of the overall current value similarity sequence. The greater the data correlation strength, the higher the correlation between the data of the column vector elements in the current similarity matrix, the stronger the ability of the corresponding current value prediction sequence to capture the trend of historical current value changes, and the greater the corresponding data correlation degree.

[0073] In a specific implementation of the embodiment of the present invention, the method for obtaining the degree of data association is expressed in the formula as follows: ;in, predicting the data association strength of the sequence for the current value; is the number of column vectors in the current value similarity matrix, that is, the number of columns; is the first column vector; is the reference similarity sequence; Calculate the cosine similarity function; is the first The cosine similarity between the column vector and the reference similarity sequence; To normalize the function, the normalization method uses linear normalization.

[0074] Step S103: weighting the priori smoothing coefficient in the exponential moving average algorithm with the data association strength to determine the adaptive smoothing coefficient of the current value prediction sequence; predicting the current value by combining the adaptive smoothing coefficient with the exponential moving average algorithm to determine the real-time current prediction data value.

[0075] The greater the data association strength, the stronger the ability of the current value prediction sequence to capture the trend of historical current value changes; then according to the properties of the smoothing coefficient in the exponential moving average algorithm, the greater the data association strength corresponding to the current value prediction sequence used for prediction, the larger the corresponding smoothing coefficient should be. Therefore, the priori smoothing coefficient in the exponential moving average algorithm is weighted by the data association strength, so that the obtained adaptive smoothing coefficient is positively correlated with the data association strength. Further, the current value is predicted based on the more accurate adaptive smoothing coefficient combined with the exponential moving average algorithm, so that the obtained real-time current prediction data value is more accurate.

[0076] Preferably, in some possible implementations of the embodiments of the present invention, the method for obtaining the adaptive smoothing coefficient in step S103 includes:

[0077] According to the product of the data association strength and the prior smoothing coefficient, the corresponding adaptive smoothing coefficient is determined. In a specific implementation of an embodiment of the present invention, the prior smoothing coefficient is set to 0.9, and the implementer can adjust it according to the specific implementation environment. For the exponential moving average algorithm, the corresponding smoothing coefficient ranges from 0 to 1. Since the data association strength ranges from 0 to 1, and the larger the prior smoothing coefficient, the more similar the corresponding predicted data is to the trend in the historical data, the stronger the ability of the corresponding current value prediction sequence to capture the trend of historical current value changes, the larger the corresponding smoothing coefficient should be; therefore, a larger prior smoothing coefficient is selected, and weighted by the data association strength, so that the obtained adaptive smoothing coefficient is more accurate.

[0078] In a specific implementation of the embodiment of the present invention, the process of obtaining the adaptive smoothing coefficient is expressed by the formula: ;in, is the adaptive smoothing coefficient of the current value prediction series; is the prior smoothing coefficient; Data association strength for the current value prediction sequence.

[0079] Preferably, in some possible implementations of the embodiments of the present invention, the method for acquiring the real-time current prediction data value in step S103 includes:

[0080] After replacing the smoothing coefficient in the exponential moving average algorithm with the adaptive smoothing coefficient, the current value prediction sequence is used as the input of the exponential moving average algorithm to output the real-time current prediction data value at the current moment. Because the larger the adaptive smoothing coefficient, the stronger the ability of the corresponding current prediction sequence to combine the historical current value change trend, and the greater the influence of the corresponding current prediction sequence on the prediction result, when the adaptive smoothing coefficient is used as the smoothing coefficient in the exponential moving average algorithm, the current value prediction sequence needs to be used as the input of the exponential moving average algorithm, so that the real-time current prediction data value obtained at the current moment is more accurate. It should be noted that the exponential moving average algorithm is a technical means well known to those skilled in the art, and will not be further elaborated here.

[0081] Step S104: performing semiconductor laser power control according to the real-time current prediction data value.

[0082] Through steps S101 to S103, a more accurate real-time current prediction data value corresponding to the current moment is predicted, that is, the current value that needs to be adjusted at the current moment, so that the current is adjusted according to the real-time current prediction data value, thereby indirectly performing more accurate semiconductor laser power control, making the semiconductor laser power more stable.

[0083] Preferably, in some possible implementations of the embodiments of the present invention, the method for controlling the power of a semiconductor laser according to the real-time current prediction data value in step S104 includes:

[0084] The difference between the real-time current prediction data value and the current value at the current moment is used as the current signal prediction deviation; the ratio between the current signal prediction deviation and the real-time current prediction data value is used as the current signal fluctuation reference value. The larger the current signal fluctuation reference value, the greater the difference between the real-time current prediction data value and the current value at the current moment, and the greater the necessity to adjust the current value at the current moment. Therefore, when the current signal fluctuation reference value is greater than the preset fluctuation threshold, the power of the semiconductor laser is controlled by adjusting the current value of the semiconductor laser to the real-time current prediction data value; when the current signal fluctuation reference value is less than or equal to the preset fluctuation threshold, the current value of the semiconductor laser is not adjusted. That is, when the real-time current prediction data value is close to the current value at the current moment, it is considered that the real-time current prediction data value is an acceptable data value in terms of current value fluctuation, so the current value of the semiconductor laser is not adjusted, otherwise the current value of the semiconductor laser is adjusted to the real-time current prediction data value, so that the power control of the semiconductor laser is more accurate. By judging whether to adjust the current value at the current moment to the real-time current prediction data value, unnecessary current adjustment processes can be reduced, complexity can be reduced, and the power control effect of the semiconductor laser can be better.

[0085] Preferably, in some possible implementations of the embodiments of the present invention, the method of adjusting the current value of the semiconductor laser to the real-time current prediction data value includes: feeding back the real-time current prediction data value to the power control module of the semiconductor laser, and adjusting the current value of the semiconductor laser to the real-time current prediction data value by the power control module through the pulse width modulation (PWM) technology. It should be noted that the pulse width modulation technology is a prior art well known to those skilled in the art and will not be further described here.

[0086] In summary, a power control method for a semiconductor laser screens out a current value similarity sequence by similarity of current value distribution between a current value prediction sequence and each current value reference sequence; thereby determining the data correlation strength of each current data corresponding to the current value prediction sequence according to the correlation between each current value similarity sequence and the reference similarity, thereby obtaining a more accurate adaptive smoothing coefficient corresponding to the exponential moving average algorithm according to the data correlation strength, so that when the current value is predicted by the adaptive smoothing coefficient combined with the exponential moving average algorithm, the real-time current prediction data value obtained is more accurate; correspondingly, the accuracy of semiconductor laser power control based on the predicted real-time current prediction data value is higher.

[0087] This application also provides a power control system for a semiconductor laser. Figure 2 , which shows a structural diagram of a power control system of a semiconductor laser provided by an embodiment of the present invention, the system includes: a data acquisition preprocessing module 201, a first determination module 202, a second determination module 203 and a semiconductor laser power control module 204.

[0088] The data acquisition preprocessing module 201 is used to acquire a current value prediction sequence representing neighborhood data at the current moment and a current value base sequence representing historical data of the semiconductor laser in the working state; traverse the current value base sequence through a sliding window with the same length as the current value prediction sequence to obtain at least two current value reference sequences;

[0089] The first determination module 202 is used to determine the reference similarity between each current value reference sequence and the current value sequence in the dimension of the current value change trend; filter out the current value similarity sequences in all the current value reference sequences according to the reference similarity; and determine the data association strength of the current value prediction sequence according to the association between the reference similarity corresponding to each current value similarity sequence and each current value similarity sequence;

[0090] The second determination module 203 is used to weight the priori smoothing coefficient in the exponential moving average algorithm with the data association strength to determine the adaptive smoothing coefficient of the current value prediction sequence; and determine the real-time current prediction data value of the current value prediction sequence by combining the adaptive smoothing coefficient with the exponential moving average algorithm;

[0091] The semiconductor laser power control module 204 is used to control the semiconductor laser power according to the real-time current prediction data value.

[0092] It should be noted that the system provided in the above embodiment is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, the power control system of a semiconductor laser and the power control method of a semiconductor laser provided in the above embodiment belong to the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0093] The present application also provides a computer device. Figure 3 , which shows a schematic diagram of the structure of a computer device provided by an embodiment of the present invention, the computer device includes a memory 301, a processor 302, and a computer program 303 stored in the memory 301 and running on the processor 302, wherein when the processor 302 executes the computer program 303, the computer device can execute any one of the power control methods for semiconductor lasers introduced above.

[0094] The embodiment of the present application also provides a computer program product. When the computer program product is run on a computer device, the computer device can execute any one of the semiconductor laser power control methods described above.

[0095] The embodiment of the present application also provides a computer-readable storage medium, in which a computer program code is stored. When the computer program code is executed on a computer device, the computer device can execute any one of the semiconductor laser power control methods described above.

[0096] In the embodiments provided in the present application, it should be understood that the provided computer device, computer program product and computer-readable storage medium are all used to execute the corresponding methods provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the methods provided above and will not be repeated here.

[0097] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0098] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A power control method for a semiconductor laser, It is characterized in that The method comprises: Collecting a current value prediction sequence representing neighborhood data at the current moment and a current value base sequence representing historical data of the semiconductor laser in a working state; traversing the current value base sequence through a sliding window with the same length as the current value prediction sequence to obtain at least two current value reference sequences; Determine the reference similarity between each current value reference sequence and the current value prediction sequence in the dimension of the current value change trend; filter out the current value similarity sequences according to the reference similarity; determine the data association strength of the current value prediction sequence according to the association between the reference similarity corresponding to each current value similarity sequence and each current value similarity sequence; The priori smoothing coefficient in the exponential moving average algorithm is weighted by the data association strength to determine the adaptive smoothing coefficient of the current value prediction sequence; the current value is predicted by combining the adaptive smoothing coefficient with the exponential moving average algorithm to determine the real-time current prediction data value; Performing semiconductor laser power control according to the real-time current prediction data value; The method for obtaining the reference similarity includes: Reversely arrange the current values ​​in each current value reference sequence to obtain a corresponding current value reverse sequence; taking each current value reference sequence and each current value reverse sequence as a target sequence in turn; The difference between the current value at each index value in the target sequence and the current value at the same index value in the current value prediction sequence is used as a reference predicted current difference corresponding to each index value in the target sequence; The difference between the mean of the reference predicted current differences corresponding to all index values ​​in the target sequence and the reference predicted current difference corresponding to each index value is used as the predicted difference standard deviation corresponding to each index value in the target sequence; Performing negative correlation normalization on the accumulated values ​​of the prediction difference standard deviations corresponding to all index values ​​in the target sequence to determine the initial similarity between the target sequence and the current value prediction sequence; The maximum value of the initial similarity between each current value reference sequence and the corresponding current value reverse sequence is used as the reference similarity between each current value reference sequence and the current value prediction sequence; The method for obtaining the similar sequence of current values ​​includes: The current value reference sequence whose corresponding reference similarity is greater than a preset similarity threshold is used as the current value screening sequence; When the initial similarity of the current value screening sequence is less than the initial similarity of the corresponding current value reverse sequence, taking the current value reverse sequence corresponding to the current value screening sequence as the current value similarity sequence; When the initial similarity of the current value screening sequence is greater than or equal to the initial similarity of the corresponding current value reverse sequence, the current value screening sequence is used as a current value similarity sequence; The method for obtaining the data association strength includes: Arrange the reference similarities of all current value similarity sequences in the order in which the corresponding current value reference sequences are obtained to obtain a reference similarity sequence; Taking each current value similarity sequence as a row of the matrix, a current value similarity matrix is ​​constructed according to all current value similarity sequences; Obtaining a column vector composed of elements in each column of the current value similarity matrix; using the cosine similarity between each column vector and the reference similarity sequence as the local correlation degree of each column vector; The normalized value of the mean of the local correlation degree of all column vectors is used as the data correlation strength of the current value prediction sequence; The method for controlling the power of a semiconductor laser according to the real-time current prediction data value comprises: The difference between the real-time current prediction data value and the current value at the current moment is used as the current signal prediction deviation; Using the ratio between the current signal prediction deviation and the real-time current prediction data value as a current signal fluctuation reference value; When the current signal fluctuation reference value is greater than a preset fluctuation threshold, the power of the semiconductor laser is controlled by adjusting the current value of the semiconductor laser to the real-time current prediction data value; When the current signal fluctuation reference value is less than or equal to the preset fluctuation threshold value, the current value of the semiconductor laser is not adjusted.

2. A power control method for a semiconductor laser according to claim 1, It is characterized in that The method for obtaining the adaptive smoothing coefficient includes: The corresponding adaptive smoothing coefficient is determined according to the product of the data association strength and the priori smoothing coefficient.

3. The power control method of a semiconductor laser according to claim 1, It is characterized in that The method for obtaining the real-time current prediction data value includes: After the smoothing coefficient in the exponential moving average algorithm is replaced by the adaptive smoothing coefficient, the current value prediction sequence is used as the input of the exponential moving average algorithm to output the real-time current prediction data value at the current moment.

4. The power control method of a semiconductor laser according to claim 1, It is characterized in that The method of adjusting the current value of the semiconductor laser to the real-time current prediction data value comprises: The real-time current prediction data value is fed back to the power control module of the semiconductor laser, so that the power control module adjusts the current value of the semiconductor laser to the real-time current prediction data value through the pulse width modulation technology.

5. A power control system for a semiconductor laser, It is characterized in that The system comprises: A data acquisition preprocessing module is used to acquire a current value prediction sequence representing neighborhood data at the current moment and a current value base sequence representing historical data of the semiconductor laser in a working state; traverse the current value base sequence through a sliding window with the same length as the current value prediction sequence to obtain at least two current value reference sequences; A first determination module is used to determine the reference similarity between each current value reference sequence and the current value sequence in the dimension of the current value change trend; filter out the current value similarity sequences in all the current value reference sequences according to the reference similarity; and determine the data association strength of the current value prediction sequence according to the association between the reference similarity corresponding to each current value similarity sequence and each current value similarity sequence; The method for obtaining the reference similarity includes: Reversely arrange the current values ​​in each current value reference sequence to obtain a corresponding current value reverse sequence; taking each current value reference sequence and each current value reverse sequence as a target sequence in turn; The difference between the current value at each index value in the target sequence and the current value at the same index value in the current value prediction sequence is used as a reference predicted current difference corresponding to each index value in the target sequence; The difference between the mean of the reference predicted current differences corresponding to all index values ​​in the target sequence and the reference predicted current difference corresponding to each index value is used as the predicted difference standard deviation corresponding to each index value in the target sequence; Performing negative correlation normalization on the accumulated values ​​of the prediction difference standard deviations corresponding to all index values ​​in the target sequence to determine the initial similarity between the target sequence and the current value prediction sequence; The maximum value of the initial similarity between each current value reference sequence and the corresponding current value reverse sequence is used as the reference similarity between each current value reference sequence and the current value prediction sequence; The method for obtaining the similar sequence of current values ​​includes: The current value reference sequence whose corresponding reference similarity is greater than a preset similarity threshold is used as the current value screening sequence; When the initial similarity of the current value screening sequence is less than the initial similarity of the corresponding current value reverse sequence, taking the current value reverse sequence corresponding to the current value screening sequence as the current value similarity sequence; When the initial similarity of the current value screening sequence is greater than or equal to the initial similarity of the corresponding current value reverse sequence, the current value screening sequence is used as a current value similarity sequence; The method for obtaining the data association strength includes: Arrange the reference similarities of all current value similarity sequences in the order in which the corresponding current value reference sequences are obtained to obtain a reference similarity sequence; Taking each current value similarity sequence as a row of the matrix, a current value similarity matrix is ​​constructed according to all current value similarity sequences; Obtaining a column vector composed of elements in each column of the current value similarity matrix; using the cosine similarity between each column vector and the reference similarity sequence as the local correlation degree of each column vector; The normalized value of the mean of the local correlation degree of all column vectors is used as the data correlation strength of the current value prediction sequence; The second determination module is used to weight the priori smoothing coefficient in the exponential moving average algorithm with the data association strength to determine the adaptive smoothing coefficient of the current value prediction sequence; and determine the real-time current prediction data value of the current value prediction sequence by combining the adaptive smoothing coefficient with the exponential moving average algorithm; A semiconductor laser power control module, used for performing semiconductor laser power control according to the real-time current prediction data value; The method for controlling the power of a semiconductor laser according to the real-time current prediction data value comprises: The difference between the real-time current prediction data value and the current value at the current moment is used as the current signal prediction deviation; Using the ratio between the current signal prediction deviation and the real-time current prediction data value as a current signal fluctuation reference value; When the current signal fluctuation reference value is greater than a preset fluctuation threshold, the power of the semiconductor laser is controlled by adjusting the current value of the semiconductor laser to the real-time current prediction data value; When the current signal fluctuation reference value is less than or equal to the preset fluctuation threshold value, the current value of the semiconductor laser is not adjusted.

6. A computer device, It is characterized in that include: A memory and a processor; wherein the memory is used to store computer program code that can be executed on the processor; The processor is used to call and run the computer program code from the memory to implement the steps of a power control method for a semiconductor laser as claimed in any one of claims 1 to 4.