CN114963630B - System debugging method of semiconductor refrigeration equipment - Google Patents
System debugging method of semiconductor refrigeration equipment Download PDFInfo
- Publication number
- CN114963630B CN114963630B CN202110210452.2A CN202110210452A CN114963630B CN 114963630 B CN114963630 B CN 114963630B CN 202110210452 A CN202110210452 A CN 202110210452A CN 114963630 B CN114963630 B CN 114963630B
- Authority
- CN
- China
- Prior art keywords
- individual
- probability
- individuals
- feature
- variation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F25—REFRIGERATION OR COOLING; COMBINED HEATING AND REFRIGERATION SYSTEMS; HEAT PUMP SYSTEMS; MANUFACTURE OR STORAGE OF ICE; LIQUEFACTION SOLIDIFICATION OF GASES
- F25B—REFRIGERATION MACHINES, PLANTS OR SYSTEMS; COMBINED HEATING AND REFRIGERATION SYSTEMS; HEAT PUMP SYSTEMS
- F25B21/00—Machines, plants or systems, using electric or magnetic effects
- F25B21/02—Machines, plants or systems, using electric or magnetic effects using Peltier effect; using Nernst-Ettinghausen effect
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F25—REFRIGERATION OR COOLING; COMBINED HEATING AND REFRIGERATION SYSTEMS; HEAT PUMP SYSTEMS; MANUFACTURE OR STORAGE OF ICE; LIQUEFACTION SOLIDIFICATION OF GASES
- F25B—REFRIGERATION MACHINES, PLANTS OR SYSTEMS; COMBINED HEATING AND REFRIGERATION SYSTEMS; HEAT PUMP SYSTEMS
- F25B2321/00—Details of machines, plants or systems, using electric or magnetic effects
- F25B2321/02—Details of machines, plants or systems, using electric or magnetic effects using Peltier effects; using Nernst-Ettinghausen effects
- F25B2321/021—Control thereof
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P80/00—Climate change mitigation technologies for sector-wide applications
- Y02P80/10—Efficient use of energy, e.g. using compressed air or pressurized fluid as energy carrier
Abstract
The invention relates to a system debugging method of semiconductor refrigeration equipment, which comprises the following steps: an initial experimental data acquisition step, namely acquiring initial experimental data by carrying out random experiments on the semiconductor refrigeration equipment for a plurality of times, wherein input data in the initial experimental data form an initial group; a fitness calculating step of calculating fitness of each individual in the population based on the experimental data; a next generation group acquisition step of performing genetic operation on the group according to probability and obtaining a next generation group if the adaptability of all individuals in the group is judged to not meet the stop condition; a new experimental data acquisition step of obtaining output data corresponding to each individual in the next generation population through an actual experimental test to form new experimental data; and returning to and circularly executing the fitness calculating step, the next generation population acquiring step and the new experimental data acquiring step until the stopping condition is met. The system debugging method has the advantages that the test result is more spectrum-dependent, and the accuracy is higher.
Description
Technical Field
The invention relates to a semiconductor refrigeration technology, in particular to a system debugging method of semiconductor refrigeration equipment.
Background
Users who use semiconductor refrigeration equipment, particularly high-capacity semiconductor refrigeration cabinets, are often high-end consumer groups who themselves have high demands on the equipment, and such equipment is also a product code for high-end high-tech on-line, and therefore, precise control of the equipment is required.
The semiconductor refrigeration equipment is precisely controlled, and more undetermined control parameters are inevitably generated. The system debugging personnel work to adjust the undetermined control parameters through experience and rules so as to achieve the expected control effect. Traditional experience-based, rule-based system debugging is generally the following procedure:
(1) Setting a set of preset parameters according to experience or rules;
(2) The operation adopts a control method of preset parameters to obtain a result;
(3) If the rule is not met, a modification scheme is formulated according to the operation result and combining personal experience and a default rule;
(4) The control method after the modification scheme is adopted is operated to obtain a result;
(5) And (3) if the flow of the step (3) and the flow of the step (4) are not in accordance with the requirements, and the flow of the step (3) and the flow of the step (4) are circularly carried out until the requirements are met.
The traditional system debugging method has the following defects:
(1) Under the condition of more undetermined parameters, the working efficiency of system debugging personnel is low, and the occupied laboratory time and the material cost are higher;
(2) The work of system debugging personnel is more and experience is hooked, so that the debugging result is changed more along with personnel;
(3) The system debugging personnel are difficult to ensure that the better undetermined control parameters are found, the debugging effect is poor, the full potential of hardware and software is difficult to develop, and the debugging has a certain probability of failure;
(4) Because the debugging method is set manually, the automatic system debugging is difficult to realize.
Disclosure of Invention
The invention aims to overcome at least one defect of the prior art and provides a system debugging method which is suitable for high-quality optimizing of data of a data set with smaller data size and larger experimental cost, and has high accuracy and high efficiency.
A further object of the present invention is to better weigh the accuracy and time costs of the optimization.
Another further object of the present invention is to improve the efficiency and accuracy of genetic algorithm optimization.
In order to achieve at least one of the above objects, the present invention provides a system debugging method of a semiconductor refrigeration apparatus, the system debugging method comprising:
an initial experimental data acquisition step, namely acquiring initial experimental data by carrying out multiple random experiments on the semiconductor refrigeration equipment, wherein input data in the initial experimental data form an initial group, all input data of each random experiment form an individual of the initial group, and different input data of the same random experiment form different characteristics of the same individual;
a fitness calculating step of calculating fitness of each individual in the population based on the experimental data;
a next generation group acquisition step of performing genetic operation on the group according to probability and obtaining a next generation group if the adaptability of all individuals in the group is judged to not meet the stop condition;
a new experimental data acquisition step of obtaining output data corresponding to each individual in the next generation population through actual experimental test to form new experimental data;
and returning and circularly executing the fitness calculating step, the next generation population acquiring step and the new experimental data acquiring step until the stopping condition is met.
Optionally, the step of obtaining output data corresponding to each individual in the next generation population by actual trial testing comprises:
and inputting each individual in the next generation group into the semiconductor refrigeration equipment successively, operating the semiconductor refrigeration equipment after inputting the semiconductor refrigeration equipment by each individual, and measuring the value of each output parameter used for representing the performance of the semiconductor refrigeration equipment so as to form output data corresponding to the individual.
Optionally, the semiconductor refrigeration device has a plurality of output parameters for characterizing its performance, and in the fitness calculating step, the fitness of the individual is calculated by the following formula:
g(Y)＝g(y 1 ，y 2 ……，y a )＝k 1 g s (y 1 )+k 2 g s (y 2 )+……+k n g s (y a ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
g s (y)＝(y-y min )/(y max -y min ) The method comprises the steps of carrying out a first treatment on the surface of the And is also provided with
g (Y) represents the fitness of the individual, Y 1 ，y 2 ……，y a Respectively representing a different output parameters; k (k) 1 ，k 2 ……，k a Weight coefficients, g, respectively representing a different output parameters s (y) is a normalization function, y min Y is the minimum threshold of the output parameter max Y is the value of the output parameter, which is the maximum threshold value of the output parameter.
Optionally, in the next generation population acquisition step, the step of performing genetic manipulation on the population according to the probability and obtaining the next generation population includes:
screening a plurality of excellent individuals from the population based on individual fitness to form an excellent individual set;
performing prior preference mutation on a first preset percentage of excellent individuals in the excellent individual set, performing random mutation on a second preset percentage of excellent individuals in the excellent individual set, performing prior preference crossover on a third preset percentage of excellent individuals in the excellent individual set, and performing random crossover on a fourth preset percentage of excellent individuals in the excellent individual set to obtain new individuals;
the new individuals obtained are combined to form the next generation population.
Optionally, the step of performing a priori preference variation on a first predetermined percentage of the excellent individuals in the set of excellent individuals comprises:
calculating the probability of each feature occurrence of an individual in the excellent individual set to obtain probability distribution of each feature and variance thereof;
determining the probability of variation of each feature according to the variance of each feature;
randomly sampling n characteristics of an individual for w times based on the probability of variation of each characteristic to obtain characteristics of the individual requiring variation, wherein w is less than or equal to n;
randomly sampling once according to probability distribution of the feature to be mutated to obtain a mutated value of the feature;
and obtaining a new mutated individual after the values of all the features of the individual which need mutation are determined.
Optionally, the step of determining the probability of each feature mutating based on the variance of each feature comprises:
the probability of each feature mutating is determined as follows:
R(x 1 )＝K 1 ×σx 1 /Σ(σx 1 +......+σx n ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
R(x 1 ) Is the characteristic x 1 Probability of occurrence of variation; k (K) 1 Adjusting the coefficient for the prior variation probability, and K 1 More than zero and less than 1; sigma x 1 、……σx n Variance of n features of the individual, respectively.
Optionally, the step of randomly mutating a second predetermined percentage of the excellent individuals in the set of excellent individuals comprises:
determining the probability of variation of each feature of an individual in the excellent individual set, wherein the probability of variation of each feature is the same;
randomly sampling n characteristics of an individual for w times based on the probability of variation of each characteristic to obtain the characteristic of the individual needing variation, wherein w is less than or equal to n;
each of the characteristics of the individual requiring variation is mutated in the following mutation manner to obtain a value for each of the characteristics after mutation:
New x＝r×(x max -x min )+x min the method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
New x is the value after variation of the feature x; r is any number between 0 and 1, x max Is the maximum threshold of feature x, x min A minimum threshold value for feature x;
and obtaining a new mutated individual after the values of all the features of the individual which need mutation are determined.
Optionally, the step of determining the probability of variation of each feature of an individual in the set of excellent individuals comprises:
the probability of each feature mutating is determined as follows:
R(x 1 )＝1/n*K 2 the method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
R(x 1 ) Is the characteristic x 1 Probability of occurrence of variation; k (K) 2 Adjust the coefficient for random variation probability, and K 2 More than zero and less than 1; n is the total number of features of the individual.
Optionally, the step of performing a priori preference crossover on a third predetermined percentage of the excellent individuals in the set of excellent individuals comprises:
randomly sampling the third preset percentage of excellent individuals m times, wherein each random sampling randomly acquires two excellent individuals to form a plurality of different individual groups, and m is the number of the third preset percentage of excellent individuals;
calculating the probability of each feature occurrence of an individual in the excellent individual set to obtain probability distribution of each feature and variance thereof;
determining the probability of each feature crossing according to the variance of each feature;
randomly sampling n characteristics of an individual for w times based on the probability of each characteristic crossing so as to obtain characteristics of the individual needing crossing;
the features of each of the two individuals in each of the groups of individuals that need to be intersected are intersected in the following manner to obtain a value after each feature intersection:
New x 11 Features x that require crossing for one of the individuals in the group of individuals 11 Features x that need to be crossed with another individual in the group of individuals 12 Values after crossing; p (X11) and P (X12) are features X which one individual needs to cross 11 Features x that need to be crossed with another individual in the group of individuals 12 The probability of occurrence;
and obtaining the crossed individual after the values of all the characteristics of the individual needing to be crossed are determined.
Optionally, the step of determining the probability of each feature crossing based on the variance of each feature comprises:
the probability of each feature crossing is determined as follows:
R(x 1 )＝K 3 ×σx 1 /Σ(σx 1 +......+σx n ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
R(x 1 ) Is the characteristic x 1 Probability of occurrence of variation; k (K) 3 Adjust the coefficients for a priori crossover probabilities, and K 3 More than zero and less than 1; sigma x 1 、……σx n Variance of n features of the individual, respectively.
Optionally, the step of randomly crossing a fourth predetermined percentage of the excellent individuals in the set of excellent individuals comprises:
c times of random sampling is carried out on the excellent individuals with the fourth preset percentage, and two excellent individuals are randomly collected in each random sampling to form a plurality of different individual groups, wherein c is the number of the excellent individuals with the fourth preset percentage;
determining the probability of each feature of an individual in the excellent set of individuals crossing, wherein the probability of each feature crossing is the same;
randomly sampling n characteristics of an individual for w times based on the probability of each characteristic crossing so as to obtain the characteristics of the individual needing crossing, wherein w is less than or equal to n;
the features of each of the two individuals in each of the groups of individuals that need to be intersected are intersected in the following manner to obtain a value after each feature intersection:
New x 11 ＝s×x 11 +(1-s)×x 12 the method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
New x 11 Features x that require crossing for one of the individuals in the group of individuals 11 Features x that need to be crossed with another individual in the group of individuals 12 The value of the corresponding characteristic after crossing, s is any number between 0 and 1;
and obtaining the crossed individual after the values of all the characteristics of the individual needing to be crossed are determined.
Optionally, the step of determining a probability of each feature of an individual in the set of excellent individuals crossing comprises:
the probability of each feature crossing is determined as follows:
R(x 1 )＝1/n×K 4 the method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
R(x 1 ) Is the characteristic x 1 Probability of occurrence of variation; k (K) 4 Adjust the coefficients for random crossover probabilities, and K 4 More than zero and less than 1; n is the total number of features of the individual.
Optionally, when the fitness calculating step and the next generation group acquiring step are performed in return, the groups in the fitness calculating step and the next generation group acquiring step are new groups formed by the initial group and the calendar group together.
Optionally, before the fitness calculating step, the system debugging method further includes:
a maximum evolution algebra setting step for setting a maximum evolution algebra; and is also provided with
The stopping condition is that the algebra of the population reaches the maximum algebra or the adaptability of at least one individual meets the preset requirement.
In practical application, the number of tests performed randomly is limited, so that the acquired data volume is smaller, and the cost of the tests in the aspects of manpower, time, equipment and the like is larger. The existing genetic algorithm needs the assistance of a fitting or interpolation model, has the defect of a heuristic optimization method aiming at a certain specific function, and makes the existing genetic algorithm difficult to perform high-quality optimization on data set data with smaller data size and larger experimental cost.
According to the method, on the one hand, the multiple output parameters of the semiconductor refrigeration equipment are optimized simultaneously based on the improved genetic algorithm, on the other hand, the method can be realized by compiling an automatic debugging script at a computer end, manual adjustment of debugging personnel is not needed, the debugging personnel are liberated, the degree of automation is high, and the debugging efficiency is high; on the other hand, the mode of gradually approaching and finally determining the optimal parameters through circularly carrying out a plurality of genetic operations and a plurality of actual test tests is very suitable for high-quality optimization of data set data with smaller data size and larger experimental cost, the test result is more spectral and higher in accuracy, the actual application scene of a genetic algorithm is expanded, and the method has better heuristic significance on similar experimental parameter optimization scenes.
The method is simple and convenient for solving the problem of multi-objective optimization which is difficult to process, the multi-objective optimization object is converted into a single-objective optimization object through a conversion function, the dimension sizes of different optimization objects are considered, and the weight duty ratio of the different modifiable optimization objects is designed, so that good balance is achieved in optimizing accuracy and time cost.
The method further improves the crossing or mutation mode of the traditional genetic algorithm aiming at excellent individuals, adds the crossing or mutation mode based on prior preference, takes the distribution rule of the existing data set into consideration on the premise of not losing the optimizing randomness, and can improve the optimizing efficiency and accuracy of the genetic algorithm to a certain extent.
The above, as well as additional objectives, advantages, and features of the present invention will become apparent to those skilled in the art from the following detailed description of a specific embodiment of the present invention when read in conjunction with the accompanying drawings.
Drawings
Some specific embodiments of the invention will be described in detail hereinafter by way of example and not by way of limitation with reference to the accompanying drawings. The same reference numbers will be used throughout the drawings to refer to the same or like parts or portions. It will be appreciated by those skilled in the art that the drawings are not necessarily drawn to scale. In the accompanying drawings:
fig. 1 is a schematic flow chart of a system debugging method of a semiconductor refrigeration device according to an embodiment of the present invention;
figure 2 is a schematic flow diagram of genetic manipulation of ethnic groups according to probability and resulting in next generation ethnic groups, according to one embodiment of the present invention;
FIG. 3 is a schematic flow chart of a priori preference variation for a first predetermined percentage of excellent individuals in a set of excellent individuals in accordance with one embodiment of the invention;
FIG. 4 is a schematic flow chart of performing random variation on a second predetermined percentage of excellent individuals in a set of excellent individuals according to one embodiment of the present invention;
FIG. 5 is a schematic flow chart diagram of a priori preference crossover for a third predetermined percentage of excellent individuals in a set of excellent individuals in accordance with one embodiment of the present invention;
FIG. 6 is a schematic flow chart diagram of random crossing of a fourth predetermined percentage of excellent individuals in a set of excellent individuals in accordance with one embodiment of the present invention.
Detailed Description
The invention provides a system debugging method of a semiconductor refrigeration device, which is provided with a storage compartment for storing articles, and utilizes a semiconductor refrigeration piece to perform refrigeration so as to adjust the temperature of the storage compartment. Because of the need for precise control of the semiconductor refrigeration equipment, the semiconductor refrigeration equipment has a plurality of output parameters that characterize its performance.
In particular, the output parameters of the semiconductor refrigeration device may include power P, noise value V, and chamber temperature rate of change
And the second derivative of the chamber temperature rate of change +.>
Etc. In general, the smaller the power P, the better, the smaller the power P, the more power-saving; the smaller the noise value V is, the better the noise value V is, the smaller the noise pollution is, and the better the user experience is; room temperature change rate->
The larger and the better the compartment temperature change rate +.>
The larger the cooling efficiency is, the higher the cooling effect is in a short time; second derivative of the chamber temperature change rate +.>
The smaller and even closer to zero the better the second derivative of the room temperature rate of change +.>
The closer to zero, the closer to zero the temperature decrease variation, the temperature can be decreased at a uniform speed.
In order to better optimize a plurality of output parameters of semiconductor refrigeration equipment, the invention provides a system debugging method based on an improved heritage algorithm.
Fig. 1 is a schematic flow chart of a system debugging method of a semiconductor refrigeration device according to an embodiment of the present invention. The system debugging method of the semiconductor refrigeration equipment comprises the following steps:
step S10, an initial experimental data acquisition step: obtaining initial experimental data by carrying out multiple random experiments on the semiconductor refrigeration equipment; the input data in the initial experimental data form an initial group, all the input data in each random test form an individual of the initial group, and different input data in the same random test form different characteristics of the same individual. For example, in one embodiment, the input data to the semiconductor refrigeration device may include the supply voltage to the semiconductor refrigeration chiller, the duty cycle of the cold side fan, and the duty cycle of the hot side fan at each random test. The power supply voltage of the semiconductor refrigerating sheet, the duty ratio of the cold end fan and the duty ratio of the hot end fan which are tested randomly at the same time are taken as different characteristics of the individuals to jointly form one individual of the group.
Step S20, a fitness calculating step: calculating fitness of each individual in the population based on the experimental data;
step S30, the next generation group acquisition step: if the adaptability of all the individuals in the family group is judged to not meet the stopping condition, carrying out genetic operation on the family group according to the probability and obtaining a next generation family group;
step S40, a new experimental data acquisition step: obtaining output data corresponding to each individual in the next generation population through actual test to form new experimental data;
step S50, returning to and circularly executing the fitness calculating step, the next generation population acquiring step and the new experimental data acquiring step until the stopping condition is met.
It should be noted that the random test in step S10 means a test without any experience or rule. This is because the empirical-based adjustment method to obtain the initial experimental data is equivalent to artificially adding a priori conditions to the initial experimental data and artificially adding restrictions to the distribution of the input data, which can result in a biased distribution of the initial experimental data that makes it difficult to traverse the entire search dataset.
According to the method, on the one hand, the multiple output parameters of the semiconductor refrigeration equipment are optimized simultaneously based on the improved genetic algorithm, on the other hand, the method can be realized by compiling an automatic debugging script at a computer end, manual adjustment of debugging personnel is not needed, the debugging personnel are liberated, the degree of automation is high, and the debugging efficiency is high; on the other hand, the mode of gradually approaching and finally determining the optimal parameters through circularly carrying out a plurality of genetic operations and a plurality of actual test tests is very suitable for high-quality optimization of data set data with smaller data size and larger experimental cost, the test result is more spectral and higher in accuracy, the actual application scene of a genetic algorithm is expanded, and the method has better heuristic significance on similar experimental parameter optimization scenes.
Further, when the fitness calculating step and the next generation group acquiring step are returned to be executed, the groups in the fitness calculating step and the next generation group acquiring step are new groups formed by the initial group and the calendar group together. That is, during the course of the cycle, the number of individuals in the population is increasing.
In some embodiments, in the new experimental data obtaining step of step S40, the step of obtaining the output data corresponding to each individual in the next generation population through the actual experimental test may specifically include:
each individual in the next generation group is sequentially input into the semiconductor refrigeration equipment, the semiconductor refrigeration equipment is operated after each individual is input into the semiconductor refrigeration equipment, and the values of various output parameters used for representing the performance of the semiconductor refrigeration equipment are measured to form output data corresponding to the individual. And after the output data corresponding to all individuals are obtained, new experimental data are formed together with the individuals in the group.
That is, in step S40, the actual test may be to input each characteristic of the individual in the population into the semiconductor refrigeration device, and obtain the value of the actual output parameter by operating the semiconductor refrigeration device. The output data obtained in this way is much more accurate than any fitting, simulation or prediction, and therefore the optimum can be obtained with as few genetic manipulations and actual experimental tests as possible.
Through a large number of experiments, a large amount of experimental data can be obtained, namely, a large amount of input data and output data are available. The traditional genetic algorithm is to find a proper mapping relation to map the input and the output. However, for a semiconductor refrigeration device having a plurality of output parameters, it is difficult to find a mapping relationship between an input and a plurality of outputs, and even if the accuracy is found is not high.
To this end, in some embodiments, the fitness calculating step of the present application may calculate the fitness of the individual by the following formula:
g(Y)＝g(y 1 ，y 2 ……，y a )＝k 1 g s (y 1 )+k 2 g s (y 2 )+……+k n g s (y a ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
g s (y)＝(y-y min )/(y max -y min ) The method comprises the steps of carrying out a first treatment on the surface of the And is also provided with
g (Y) represents the fitness of the individual, Y 1 ，y 2 ……，y a Respectively representing a different output parameters; k (k) 1 ，k 2 ……，k a Weight coefficients, g, respectively representing a different output parameters s (y) is a normalization function, y min Y is the minimum threshold of the output parameter max Y is the value of the output parameter, which is the maximum threshold value of the output parameter.
For example, when a is 3, and y 1 ，y 2 ，y 3 G when the three output parameters of the power supply voltage of the semiconductor refrigeration sheet, the duty ratio of the cold end fan and the duty ratio of the hot end fan are respectively expressed s (y 1 )＝(y 1 -y 1min )/(y 1max -y 1min )，g s (y 2 )＝(y 2 -y 2min )/(y 2max -y 2min )，g s (y 3 )＝(y 3 -y 3min )/(y 3max -y 3min ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein g s (y 1 )、g s (y 2 )、g s (y 3 ) Respectively normalizing the three output parameters, y 1 、y 2 、y 3 The values of the three output parameters corresponding to the individuals are respectively; y is 1max 、y 1min The maximum threshold value and the minimum threshold value of the power supply voltage of the semiconductor refrigerating sheet are respectively set; y is 2max 、y 2min The maximum threshold value and the minimum threshold value of the duty ratio of the cold end fan are respectively; y is 3max 、y 3min The maximum threshold and the minimum threshold of the duty ratio of the hot-end fan are respectively set.
Therefore, the method for optimizing the multi-objective targets is simple and convenient for processing aiming at the multi-objective optimization problem which is difficult to process, the multi-objective optimization targets are converted into single-objective optimization targets through the conversion function, the dimension sizes of different optimization targets are considered, and the weight duty ratio of the different modifiable optimization targets is designed, so that better balance is achieved in optimizing accuracy and time cost.
Figure 2 is a schematic flow diagram of genetic manipulation of ethnic groups according to probability and obtaining next generation ethnic groups, according to one embodiment of the present invention. In some embodiments, in the next generation population acquisition step, the step of performing genetic manipulation on the population according to the probability and obtaining the next generation population may specifically include:
step S31, screening a plurality of excellent individuals from the population based on the individual fitness to form an excellent individual set; specifically, the preferred individuals of the group, which are arranged in the first predetermined percentage or predetermined number, can be selected to form the excellent individual set based on the ranking of the fitness of the individuals.
Step S32, performing prior preference mutation on a first preset percentage of excellent individuals in the excellent individual set, performing random mutation on a second preset percentage of excellent individuals in the excellent individual set, performing prior preference crossover on a third preset percentage of excellent individuals in the excellent individual set, and performing random crossover on a fourth preset percentage of excellent individuals in the excellent individual set to obtain new individuals. It is understood that the first, second, third, and fourth preset percentages may be set based on predictions or experience, and the sum of the first, second, third, and fourth preset percentages is not necessarily one hundred percent, but may be less than one hundred percent, or may be greater than one hundred percent. Moreover, the excellent individuals who perform the prior preference mutation, the random mutation, the prior preference crossover and the random crossover are all randomly selected, and more than two genetic operations may be performed on some excellent individuals.
Step S33, combining the obtained new individuals to form a next generation population. The next generation population is formed for performing the actual test at step S40.
The method further improves the crossing or mutation mode of the traditional genetic algorithm aiming at excellent individuals, adds the crossing or mutation mode based on prior preference, takes the distribution rule of the existing data set into consideration on the premise of not losing the optimizing randomness, and can improve the optimizing efficiency and accuracy of the genetic algorithm to a certain extent.
In addition, the improved genetic algorithm does not particularly encode the input data, and the spatial mapping relation and the actual distribution of decimal input data are reserved.
FIG. 3 is a schematic flow chart of a priori preference variation for a first predetermined percentage of excellent individuals in a set of excellent individuals, according to one embodiment of the invention. In some embodiments, the step of performing a priori preference variation on a first predetermined percentage of the excellent individuals in the set of excellent individuals comprises:
step S3211, calculating the probability of each feature occurrence of the individuals in the excellent set of individuals to obtain a probability distribution of each feature and its variance. Specifically, for discrete features, the probability distribution and variance of each feature may be obtained by a common probability calculation method, and for continuous features, the probability distribution and variance of each feature may be obtained by a normal distribution estimation method.
Step S3212, determining the probability of variation of each feature according to the variance of each feature;
step S3213, randomly sampling n features of the individual for w times based on the probability of variation of each feature to obtain features of the individual requiring variation, wherein w is less than or equal to n; it will be appreciated that w random samples may or may not be less than w features requiring variation.
Step S3214, randomly sampling once according to the probability distribution of the feature to be mutated to obtain a mutated value of the feature;
step S3215, obtaining a new mutated individual after determining the values of all the features of the individual which need mutation.
Further, the probability of variation of different characteristics of an individual is different for individuals who have a priori preference variation. The variance of different features can reflect the probability of variation of the different features, and the smaller the variance is, the smaller the fluctuation of the features is, and the probability of variation of the features is reduced; the larger the variance, the greater the uncertainty that accounts for the feature, and the greater the probability that the feature will be mutated.
Specifically, the probability of variation of each feature can be determined as follows:
R(x 1 )＝K 1 ×σx 1 /Σ(σx 1 +......+σx n ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
R(x 1 ) Is the characteristic x 1 Probability of occurrence of variation; k (K) 1 Adjusting the coefficient for the prior variation probability, and K 1 More than zero and less than 1; sigma x 1 、……σx n Variance of n features of the individual, respectively. Preferably, K 1 The value of (2) is between 0.8 and 1.
The addition of the prior preference variation probability adjustment coefficient in the probability calculation of the feature transmission prior preference variation can obtain an optimal value with fewer actual test times.
FIG. 4 is a schematic flow chart of random variation of a second predetermined percentage of excellent individuals in a set of excellent individuals according to one embodiment of the invention. In some embodiments, the step of randomly mutating a second predetermined percentage of the excellent individuals in the set of excellent individuals comprises:
step S3221, determining the probability of variation of each feature of an individual in the excellent individual set, wherein the probability of variation of each feature is the same;
step S3222, randomly sampling n characteristics of the individual based on the probability of variation of each characteristic, so as to obtain the characteristic of the individual needing variation, wherein w is less than or equal to n; it will be appreciated that w random samples may or may not be less than w features requiring variation.
Step S3223, mutating each characteristic of the individual to be mutated in the following mutation manner to obtain a value after mutation of each characteristic:
New x＝r×(x max -x min )+x min the method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
New x is the value after variation of the feature x; r is any number between 0 and 1, x max Is the maximum threshold of feature x, x min A minimum threshold value for feature x;
and S3224, obtaining a new mutated individual after the values of all the features of the individual which need mutation are determined.
Further, the probability of variation of different characteristics of an individual is the same for individuals who have random variation. Specifically, the probability of variation of each feature can be determined as follows:
R(x 1 )＝1/n*K 2 the method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
R(x 1 ) Is the characteristic x 1 Probability of occurrence of variation; k (K) 2 Adjust the coefficient for random variation probability, and K 2 More than zero and less than 1; n is the total number of features of the individual. Preferably, K 2 Is taken to be at0.8 to 1.
Adding the random variation probability adjustment coefficient to the probability calculation of the feature transmission random variation can obtain an optimal value with fewer actual test times.
FIG. 5 is a schematic flow chart diagram of a priori preference crossover for a third predetermined percentage of excellent individuals in a set of excellent individuals in accordance with one embodiment of the present invention. In some embodiments, the step of a priori preference crossing for a third predetermined percentage of the excellent individuals in the set of excellent individuals may specifically include:
step S3231, randomly sampling the third preset percentage of excellent individuals for m times, wherein each random sampling randomly collects two excellent individuals to form a plurality of different individual groups, and m is the number of the third preset percentage of excellent individuals;
step S3232, calculating the probability of each feature of an individual in the excellent individual set to obtain probability distribution of each feature and variance thereof;
step S3233, determining the probability of each feature crossing according to the variance of each feature;
step S3234, randomly sampling n features of an individual for w times based on the probability of each feature crossing to obtain features of the individual needing crossing; it will be appreciated that w random samples may or may not be less than w features to be intersected.
Step S3235, intersecting the features that each of the two individuals in each individual group need to intersect in the following intersecting manner to obtain a value after each feature intersection:
New x 11 Features x that require crossing for one of the individuals in the group of individuals 11 Features x that need to be crossed with another individual in the group of individuals 12 Values after crossing; p (X11) and P (X12) are features X which one individual needs to cross 11 And another individual in the group of individuals needsCross feature x 12 The probability of occurrence;
and step S3236, obtaining the crossed individual after the values of all the characteristics of the individual to be crossed are determined.
It is to be understood that the above step S3231 is not limited to occur before the step S3232, and may occur before or after any one of the step S3232, the step S3233, and the step S3234, as long as the step S3231 is ensured before the step S3235.
In two genetic operations of performing a priori bias crossover and a priori bias mutation on an individual, the difference is the specific manner in which the individual transmits crossover and mutation, and the determination manner of the characteristics of the individual that need crossover or mutation is the same, so that the description thereof will not be repeated here.
Specifically, the probability of each feature crossing may be determined as follows:
R(x 1 )＝K 3 ×σx 1 /Σ(σx 1 +......+σx n ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
R(x 1 ) Is the characteristic x 1 Probability of occurrence of variation; k (K) 3 Adjust the coefficients for a priori crossover probabilities, and K 3 More than zero and less than 1; sigma x 1 、……σx n Variance of n features of the individual, respectively. Preferably, K 3 The value of (2) is between 0.8 and 1.
The addition of the prior preference crossover probability adjustment coefficient in the probability calculation of the feature transmission prior preference crossover can obtain an optimal value with fewer actual test times.
FIG. 6 is a schematic flow chart diagram of random crossing of a fourth predetermined percentage of excellent individuals in a set of excellent individuals in accordance with one embodiment of the present invention. In some embodiments, the step of randomly crossing a fourth predetermined percentage of the excellent individuals in the set of excellent individuals may specifically include:
step S3241, performing random sampling on the fourth preset percentage of excellent individuals for c times, wherein each random sampling randomly collects two excellent individuals to form a plurality of different individual groups, and c is the number of the fourth preset percentage of excellent individuals;
step S3242, determining the probability of each feature of an individual in the excellent individual set to cross, wherein the probability of each feature crossing is the same;
step S3243, randomly sampling n features of an individual for w times based on the probability of each feature crossing to obtain features of the individual needing crossing, wherein w is less than or equal to n; it will be appreciated that w random samples may or may not be less than w features to be intersected.
Step S3244, intersecting the features that each of the two individuals in each individual group need to intersect in the following intersecting manner to obtain a value after each feature intersection:
New x 11 ＝s×x 11 +(1-s)×x 12 the method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
New x 11 Features x that require crossing for one of the individuals in the group of individuals 11 Features x that need to be crossed with another individual in the group of individuals 12 The value of the corresponding characteristic after crossing, s is any number between 0 and 1;
and step S3245, obtaining the crossed individual after the values of all the characteristics of the individual to be crossed are determined.
It is to be understood that the step S3241 is not limited to occur before the step S3242, and may occur before or after any one of the step S3242 and the step S3243, as long as the step S3241 is ensured before the step S3244.
In two genetic operations of random crossing and random mutation on an individual, the specific manner in which the individual transmits the crossing and mutation is different, and the manner in which the characteristics of the individual that need to undergo the crossing or mutation are determined is the same, so that the description thereof will not be repeated here.
Specifically, the probability of each feature crossing may be determined as follows:
R(x 1 )＝1/n×K 4 the method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
R(x 1 ) Is the characteristic x 1 Probability of occurrence of variation; k (K) 4 Adjust the coefficients for random crossover probabilities, and K 4 More than zero and less than 1; n is an individualTotal number of features. Preferably, K 4 The value of (2) is between 0.8 and 1.
In some embodiments, before the fitness calculation step, the system debugging method of the present invention may further include: and setting the maximum evolutionary algebra, namely setting the maximum evolutionary algebra.
Further, the stopping condition of the present invention may be that the algebra of the population reaches the maximum algebra or that the fitness of at least one individual meets the preset requirement. The setting of the maximum evolution algebra can avoid the long-time ineffective cyclic execution of the system debugging method, and improves the debugging efficiency.
It will be appreciated by those skilled in the art that the semiconductor refrigeration device of the present invention may be a storage device utilizing a semiconductor refrigeration tablet for refrigeration, for example, the semiconductor refrigeration device of the present invention includes, but is not limited to, a semiconductor refrigeration wine cabinet, a semiconductor refrigeration refrigerator, a semiconductor refrigeration ice chest, and the like.
In some embodiments, the system debugging method of the semiconductor refrigeration equipment is particularly suitable for high-capacity semiconductor solid refrigeration wine cabinets with the capacity of more than 400L, and the wine cabinets have larger space and higher control precision requirement.
By now it should be appreciated by those skilled in the art that while a number of exemplary embodiments of the invention have been shown and described herein in detail, many other variations or modifications of the invention consistent with the principles of the invention may be directly ascertained or inferred from the present disclosure without departing from the spirit and scope of the invention. Accordingly, the scope of the present invention should be understood and deemed to cover all such other variations or modifications.
Claims (14)
1. A system debugging method of a semiconductor refrigeration device, the system debugging method comprising:
an initial experimental data acquisition step, namely acquiring initial experimental data by carrying out multiple random experiments on the semiconductor refrigeration equipment, wherein input data in the initial experimental data form an initial group, all input data of each random experiment form an individual of the initial group, and different input data of the same random experiment form different characteristics of the same individual;
a fitness calculating step of calculating fitness of each individual in the population based on the experimental data;
a next generation group acquisition step of performing genetic operation on the group according to probability and obtaining a next generation group if the adaptability of all individuals in the group is judged to not meet the stop condition;
a new experimental data acquisition step of obtaining output data corresponding to each individual in the next generation population through actual experimental test to form new experimental data;
and returning and circularly executing the fitness calculating step, the next generation population acquiring step and the new experimental data acquiring step until the stopping condition is met.
2. The system debugging method of claim 1, wherein obtaining output data corresponding to each individual in the next generation population by actual trial testing comprises:
and inputting each individual in the next generation group into the semiconductor refrigeration equipment successively, operating the semiconductor refrigeration equipment after inputting the semiconductor refrigeration equipment by each individual, and measuring the value of each output parameter used for representing the performance of the semiconductor refrigeration equipment so as to form output data corresponding to the individual.
3. The system debugging method of claim 1, wherein the semiconductor refrigeration device has a plurality of output parameters for characterizing its performance, and in the fitness calculating step, the fitness of an individual is calculated by the following formula:
g(Y)＝g(y 1 ，y 2 ……，y a )＝k 1 g s (y 1 )+k 2 g s (y 2 )+……+k n g s (y a ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
g s (y)＝(y-y min )/(y max -y min ) The method comprises the steps of carrying out a first treatment on the surface of the And is also provided with
g(Y)Indicating the fitness of the individual, y 1 ，y 2 ……，y a Respectively representing a different output parameters; k (k) 1 ，k 2 ……，k a Weight coefficients, g, respectively representing a different output parameters s (y) is a normalization function, y min Y is the minimum threshold of the output parameter max Y is the value of the output parameter, which is the maximum threshold value of the output parameter.
4. The system debugging method of claim 1, wherein in the next generation population acquisition step, the step of performing genetic manipulation on the population according to probability and obtaining the next generation population comprises:
screening a plurality of excellent individuals from the population based on individual fitness to form an excellent individual set;
performing prior preference mutation on a first preset percentage of excellent individuals in the excellent individual set, performing random mutation on a second preset percentage of excellent individuals in the excellent individual set, performing prior preference crossover on a third preset percentage of excellent individuals in the excellent individual set, and performing random crossover on a fourth preset percentage of excellent individuals in the excellent individual set to obtain new individuals;
the new individuals obtained are combined to form the next generation population.
5. The system commissioning method of claim 4, wherein the step of a priori preference variation of a first predetermined percentage of the excellent individuals in the set of excellent individuals comprises:
calculating the probability of each feature occurrence of an individual in the excellent individual set to obtain probability distribution of each feature and variance thereof;
determining the probability of variation of each feature according to the variance of each feature;
randomly sampling n characteristics of an individual for w times based on the probability of variation of each characteristic to obtain characteristics of the individual requiring variation, wherein w is less than or equal to n;
randomly sampling once according to probability distribution of the feature to be mutated to obtain a mutated value of the feature;
and obtaining a new mutated individual after the values of all the features of the individual which need mutation are determined.
6. The system debugging method of claim 5, wherein determining the probability of each feature mutating based on the variance of each feature comprises:
the probability of each feature mutating is determined as follows:
R(x 1 )＝K 1 ×σx 1 /Σ(σx 1 +......+σx n ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
R(x 1 ) Is the characteristic x 1 Probability of occurrence of variation; k (K) 1 Adjusting the coefficient for the prior variation probability, and K 1 More than zero and less than 1; sigma x 1 、……σx n Variance of n features of the individual, respectively.
7. The system commissioning method of claim 4, wherein randomly mutating a second predetermined percentage of the set of excellent individuals comprises:
determining the probability of variation of each feature of an individual in the excellent individual set, wherein the probability of variation of each feature is the same;
randomly sampling n characteristics of an individual for w times based on the probability of variation of each characteristic to obtain the characteristic of the individual needing variation, wherein w is less than or equal to n;
each of the characteristics of the individual requiring variation is mutated in the following mutation manner to obtain a value for each of the characteristics after mutation:
New x＝r×(x max -x min )+x min the method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
New x is the value after variation of the feature x; r is any number between 0 and 1, x max Is the maximum threshold of feature x, x min A minimum threshold value for feature x;
and obtaining a new mutated individual after the values of all the features of the individual which need mutation are determined.
8. The system commissioning method of claim 7, wherein determining a probability of variation for each feature of an individual in the set of excellent individuals comprises:
the probability of each feature mutating is determined as follows:
R(x 1 )＝1/n*K 2 the method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
R(x 1 ) Is the characteristic x 1 Probability of occurrence of variation; k (K) 2 Adjust the coefficient for random variation probability, and K 2 More than zero and less than 1; n is the total number of features of the individual.
9. The system commissioning method of claim 4, wherein the step of a priori preference crossing over a third predetermined percentage of the excellent individuals in the set of excellent individuals comprises:
randomly sampling the third preset percentage of excellent individuals m times, wherein each random sampling randomly acquires two excellent individuals to form a plurality of different individual groups, and m is the number of the third preset percentage of excellent individuals;
calculating the probability of each feature occurrence of an individual in the excellent individual set to obtain probability distribution of each feature and variance thereof;
determining the probability of each feature crossing according to the variance of each feature;
randomly sampling n characteristics of an individual for w times based on the probability of each characteristic crossing so as to obtain characteristics of the individual needing crossing;
the features of each of the two individuals in each of the groups of individuals that need to be intersected are intersected in the following manner to obtain a value after each feature intersection:
New x 11 Need to be crossed for one of the individuals in the group of individualsCharacteristic x of fork 11 Features x that need to be crossed with another individual in the group of individuals 12 Values after crossing; p (X11) and P (X12) are features X which one individual needs to cross 11 Features x that need to be crossed with another individual in the group of individuals 12 The probability of occurrence;
and obtaining the crossed individual after the values of all the characteristics of the individual needing to be crossed are determined.
10. The system debugging method of claim 9, wherein determining the probability of each feature crossing based on the variance of each feature comprises:
the probability of each feature crossing is determined as follows:
R(x 1 )＝K 3 ×σx 1 /Σ(σx 1 +......+σx n ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
R(x 1 ) Is the characteristic x 1 Probability of occurrence of variation; k (K) 3 Adjust the coefficients for a priori crossover probabilities, and K 3 More than zero and less than 1; sigma x 1 、……σx n Variance of n features of the individual, respectively.
11. The system commissioning method of claim 4, wherein randomly crossing a fourth predetermined percentage of the set of excellent individuals comprises:
c times of random sampling is carried out on the excellent individuals with the fourth preset percentage, and two excellent individuals are randomly collected in each random sampling to form a plurality of different individual groups, wherein c is the number of the excellent individuals with the fourth preset percentage;
determining the probability of each feature of an individual in the excellent set of individuals crossing, wherein the probability of each feature crossing is the same;
randomly sampling n characteristics of an individual for w times based on the probability of each characteristic crossing so as to obtain the characteristics of the individual needing crossing, wherein w is less than or equal to n;
the features of each of the two individuals in each of the groups of individuals that need to be intersected are intersected in the following manner to obtain a value after each feature intersection:
New x 11 ＝s×x 11 +(1-s)×x 12 the method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
New x 11 Features x that require crossing for one of the individuals in the group of individuals 11 Features x that need to be crossed with another individual in the group of individuals 12 The value of the corresponding characteristic after crossing, s is any number between 0 and 1;
and obtaining the crossed individual after the values of all the characteristics of the individual needing to be crossed are determined.
12. The system commissioning method of claim 11, wherein determining a probability that each feature of an individual in the set of excellent individuals crosses comprises:
the probability of each feature crossing is determined as follows:
R(x 1 )＝1/n×K 4 the method comprises the steps of carrying out a first treatment on the surface of the Wherein the method comprises the steps of
R(x 1 ) Is the characteristic x 1 Probability of occurrence of variation; k (K) 4 Adjust the coefficients for random crossover probabilities, and K 4 More than zero and less than 1; n is the total number of features of the individual.
13. The system debugging method of claim 1, wherein,
when the fitness calculating step and the next generation group acquiring step are executed again, the groups in the fitness calculating step and the next generation group acquiring step are new groups formed by the initial group and the calendar group together.
14. The system debugging method of claim 1, wherein prior to the fitness computing step, the system debugging method further comprises:
a maximum evolution algebra setting step for setting a maximum evolution algebra; and is also provided with
The stopping condition is that the algebra of the population reaches the maximum algebra or the adaptability of at least one individual meets the preset requirement.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110210452.2A CN114963630B (en) | 2021-02-25 | 2021-02-25 | System debugging method of semiconductor refrigeration equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110210452.2A CN114963630B (en) | 2021-02-25 | 2021-02-25 | System debugging method of semiconductor refrigeration equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114963630A CN114963630A (en) | 2022-08-30 |
| CN114963630B true CN114963630B (en) | 2023-06-16 |
Family
ID=82974252
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110210452.2A Active CN114963630B (en) | 2021-02-25 | 2021-02-25 | System debugging method of semiconductor refrigeration equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114963630B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6230497B1 (en) * | 1999-12-06 | 2001-05-15 | Motorola, Inc. | Semiconductor circuit temperature monitoring and controlling apparatus and method |
| CN110857961A (en) * | 2018-08-22 | 2020-03-03 | 珠海格力电器股份有限公司 | Debugging method, device and system for air conditioner motor |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA2509207C (en) * | 2002-12-09 | 2012-04-24 | Hudson Technologies, Inc. | Method and apparatus for optimizing refrigeration systems |
| US7111469B2 (en) * | 2004-09-22 | 2006-09-26 | Horan Christopher J | Process for refrigerant charge level detection using a neural net |
| EP3084691A2 (en) * | 2013-12-20 | 2016-10-26 | Gemtrex Inc. | A genetic method for the acquisition and tracking of time varying signals |
-
2021
- 2021-02-25 CN CN202110210452.2A patent/CN114963630B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6230497B1 (en) * | 1999-12-06 | 2001-05-15 | Motorola, Inc. | Semiconductor circuit temperature monitoring and controlling apparatus and method |
| CN110857961A (en) * | 2018-08-22 | 2020-03-03 | 珠海格力电器股份有限公司 | Debugging method, device and system for air conditioner motor |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114963630A (en) | 2022-08-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112149209B (en) | Optimization method for multi-performance oriented design of building | |
| CN110083910B (en) | NSGA-II based chaotic time sequence prediction sample acquisition method | |
| CN111598224A (en) | Air conditioner cold load prediction method for optimizing BP neural network based on simulated annealing algorithm | |
| CN111967183A (en) | Method and system for calculating line loss of distribution network area | |
| CN112307677A (en) | Power grid oscillation mode evaluation and safety active early warning method based on deep learning | |
| Zhou et al. | Development of data-driven thermal sensation prediction model using quality-controlled databases | |
| CN111814388A (en) | CFD simulation verification method for lower air supply data center based on neural network | |
| CN114963630B (en) | System debugging method of semiconductor refrigeration equipment | |
| Gao et al. | A sequential budget allocation framework for simulation optimization | |
| CN109657695A (en) | A kind of fuzzy division clustering method and device based on definitive operation | |
| CN116955963B (en) | Heating ventilation energy-saving ladder optimizing control method based on historical data analysis | |
| CN111932066A (en) | Intelligent degree evaluation method of intelligent household product based on fuzzy analytic hierarchy process | |
| CN116400266A (en) | Transformer fault detection method, device and medium based on digital twin model | |
| CN115622034A (en) | Internet of things meter power load identification method based on improved genetic neural network | |
| Gray | Use of the Bayesian family of methods to correct for effects of exposure measurement error in polynomial regression models | |
| CN115859190A (en) | Non-invasive household electrical classification method based on causal relationship | |
| Wang et al. | Combined sensitivity ranking of input parameters and model forms of building energy simulation | |
| CN108445749B (en) | Parameter setting method applied to high-order sliding mode controller | |
| Zhu et al. | Application of machine learning and its improvement technology in modeling of total energy consumption of air conditioning water system | |
| CN111222736A (en) | Ammunition storage reliability evaluation method based on mixed correlation vector machine model | |
| CN110969340A (en) | Method and system for determining basic capability of intelligent electric energy meter quality technology | |
| CN110807599A (en) | Method, device, server and storage medium for deciding electrochemical energy storage scheme | |
| Zhu et al. | Surrogating the response PDF of stochastic simulators using generalized lambda distributions | |
| CN117236187B (en) | Parameterized design method and system for deep learning accelerator chip | |
| Zhou et al. | Optimization of Metro Central Air Conditioning Cold Source System Based on PCA-ANN Data Model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |