CN112580685A

CN112580685A - MLCC (multilayer ceramic capacitor) capacitance parameter matching method

Info

Publication number: CN112580685A
Application number: CN202011299347.2A
Authority: CN
Inventors: 郑鑫; 陈建琪; 徐楠楠
Original assignee: Qingdao Mengdou Network Technology Co ltd
Current assignee: Qingdao Mengdou Network Technology Co ltd
Priority date: 2020-11-19
Filing date: 2020-11-19
Publication date: 2021-03-30
Anticipated expiration: 2040-11-19
Also published as: CN112580685B

Abstract

The invention belongs to the technical field of artificial intelligence and electric devices, and provides an MLCC (multilayer ceramic capacitor) capacitance parameter matching method, which comprises the following steps: s1, collecting material description content input by a user; judging whether the material description has a brand and an original plant model; s2, extracting the substrings corresponding to the original factory model in a segmented manner to obtain parameters of the MLCC capacitor of the substring, and matching corresponding products and alternative products; s3, extracting parameters of the pure material description, and matching corresponding products and alternative products; s4, calculating scores of the matched corresponding products and the alternative products thereof; s5, taking each score of each product as the input of the product, and calculating the probability that the user may purchase the product by using a naive Bayes algorithm; and sorting the products with the attributes as purchased according to the probability, and generating a recommendation sequence. The invention greatly improves the efficiency and the precision of MLCC capacitance matching.

Description

MLCC (multilayer ceramic capacitor) capacitance parameter matching method

Technical Field

The invention belongs to the technical field of artificial intelligence and electric devices, and particularly relates to a matching method of MLCC capacitance parameters.

Background

Due to the relative nature of the MLCC capacitance and the general utility of the MLCC capacitance. The description method of the MLCC capacitor in the industry is various, and meanwhile, the use requirement of the MLCC capacitor in the industry is emphasized, in most cases, whether the key parameters are matched or not is emphasized, and whether the conditions of development and manufacturing can be met or not is met; some may also make a corresponding request for a brand, but not for a particular product under the brand. Due to the relevant reasons, the description of the MLCC is more variable for purchasing personnel in enterprises, and meanwhile, in some enterprises, with the industrial experience habits of purchasing personnel on the MLCC, the downward transmission of the experience of the habits and the like, the MLCC capacitor in a purchase list is often lost or not concerned with the original plant model information, but the key parameters of the MLCC capacitor are described according to the purchasing habits of the enterprises.

The diversity of the MLCC capacitor is described, so that the MLCC capacitor is difficult to identify, and the key parameter decomposition difficulty is high. In the current market, the existing decomposition and matching method for the description of the MLCC capacitor is basically lacked. For the existing decomposition matching of the MLCC capacitor, most of the MLCC capacitor products are fixed based on a description method, and can be decomposed correctly to be matched with the correct MLCC capacitor product only in a description mode meeting the requirement.

The contents are decomposed in a manual mode, the manual speed in the traditional purchasing industry is 1min per item on average, the efficiency is low, and meanwhile, the contents are influenced by factors such as manual experience, working time and the like.

Disclosure of Invention

In order to meet the actual requirements of matching of MLCC (multilayer ceramic capacitor), the invention overcomes the defects in the prior art, and the technical problem to be solved is to provide a matching method of MLCC capacitance parameters, which is used for completing the determination of categories, the decomposition of key parameters, and the confirmation and matching of brands by decomposing the material description content input by a user.

In order to solve the technical problems, the invention adopts the technical scheme that: a method for matching MLCC capacitance parameters comprises the following steps:

s1, collecting material description contents input by a user; judging whether the material description has a brand and an original factory model, if so, deleting brand information in the material description, and directly entering step S2 for parameter decomposition; if no brand exists and only the original factory model exists, reversely deducing the brand and the series corresponding to the original factory model according to the original factory model, and then entering the step S3 to carry out a parameter decomposition process; if the model of the original factory and the brand information do not exist, judging whether the material description input by the user is an MLCC capacitor or not according to the judgment, and if so, entering a step S3 to perform a parameter decomposition process;

s2, extracting the substrings corresponding to the original factory model in a segmented manner to obtain English system encapsulation of the MLCC, the MLCC material, the capacitance value, the precision and the rated voltage, and matching corresponding products and replaceable products;

s3, judging whether characters corresponding to material, encapsulation, voltage, precision and rated voltage exist in the character string corresponding to the material description content input by the user, if so, outputting corresponding parameter values, and matching corresponding products and alternative products;

s4, calculating scores of the matched corresponding products and the alternative products thereof;

s5, taking each score of each product as the input of the product, and calculating the probability that the user may purchase the product by using a naive Bayes algorithm; and sorting the products with the attributes as purchased according to the probability, and generating a recommendation sequence.

In step S1, the specific method of determining whether the material description has a brand and a genuine manufacturer model includes:

s101, judging whether a user inputs brand information or not according to the material description input by the user, if so, outputting a corresponding uniform brand name, deleting material bytes corresponding to the brand, and entering the step S102; if not, directly entering step S102;

s102, segmenting a character string corresponding to the material description without the brand according to the spacer, removing invalid characters in the segmented sub-character string, judging whether the sub-character is a combination of pure letters and numbers and whether the length is more than 8, and if so, judging the character string to be a possible original factory model; and if not, judging that the material description of the original factory model is not available.

In step S1, the specific method for determining whether the material description input by the user is the MLCC capacitor is as follows:

judging whether the name or alias of the MLCC capacitor exists in the material description input by the user, if so, judging the MLCC capacitor, and if not, entering the next step;

judging whether the specific packaging of the MLCC capacitor exists in the material description input by the user, if so, judging the MLCC capacitor, and if not, entering the next step;

judging whether the specific packaging and material combination of the MLCC capacitor exists in the material description input by the user, if so, judging the MLCC capacitor, and if not, entering the next step;

and judging whether the specific parameter combination description of the MLCC capacitor exists in the material description input by the user, if so, judging the MLCC capacitor, and if not, judging the MLCC capacitor is not the MLCC capacitor or the MLCC capacitor material description which is seriously lost.

The step S2 specifically includes the following steps:

s201, extracting characters corresponding to English system packaging character bits in a sub-character string corresponding to the original factory model, judging whether characters correspond to the characters in an English system packaging table of the MLCC, if so, outputting English system packaging of the MLCC and entering the next step, if not, deleting key parameters of the original factory model or making errors in the original factory model, and finishing decomposition;

s202, extracting characters corresponding to material character bits in a sub-character string corresponding to the original factory model, judging whether characters correspond to the characters in a material table of the MLCC, if so, outputting the materials of the MLCC and entering the next step, and if not, directly entering the next step;

s203, extracting characters corresponding to the capacitance character bits in the substring corresponding to the original factory model, judging whether three digits or a combination of two digits with a character 'R' is sandwiched, if so, calculating the capacitance, and entering the next step, otherwise, directly entering the next step;

s204, extracting characters corresponding to precision character bits in the sub-character strings corresponding to the original factory models, judging whether letters correspond to the letters in the precision table of the MLCC capacitor, if so, outputting the precision, and entering the next step, otherwise, directly entering the next step;

s205, extracting characters corresponding to rated voltage character bits in the sub-character strings corresponding to the original factory models, judging whether three digits or a combination of two digits with a character 'R' is sandwiched, and if yes, calculating a rated voltage value;

and S206, matching corresponding products and alternative products according to the English system encapsulation of the MLCC capacitor, the MLCC capacitor material, the capacitor capacitance value, the precision and the rated voltage information obtained in the steps S201 to S205.

The English system encapsulation character position in the corresponding sub character string of the original factory model is 4 th to 7 th, the material character position is 8 th to 10 th, the capacitance value character position is 11 th to 13 th, the precision character position is 14 th, and the rated voltage character position is 15 th to 17 th.

The step S3 specifically includes the following steps:

s301, extracting character strings corresponding to the precision in the material description, and replacing the character strings according to a unified standard;

s302, extracting character strings representing MLCC capacitor materials in the material description, outputting uniform material symbols, and deleting original characters representing the capacitor materials;

s303, extracting character strings representing MLCC capacitor packaging in the material description, outputting uniform capacitor packaging, and deleting the character strings representing the capacitor packaging;

s304, extracting a character string representing the MLCC capacitance precision in the material description, outputting the precision, and deleting the character string representing the capacitance precision;

s305, extracting character strings representing units in the material description, and unifying the units;

s306, extracting a character string representing voltage in the material description, outputting the voltage, and deleting the character string representing the voltage;

s307, extracting the character string representing the capacitance value in the material description, outputting the capacitance value, and deleting the character string representing the capacitance value.

In the step S304, if the precision is not successfully extracted and the character string representing the capacitance value in the material description is extracted in the step S307, the number of the numerical values in front of pf is 2, and the corresponding numerical values are all less than or equal to 10, the value with the small numerical value is used as the precision value of the MLCC capacitor;

if the voltage value and the capacitance value are not successfully extracted in the steps S306 and S307, it is determined that the material description describes the capacitance value and the rated voltage in a scientific counting method, and at this time, the voltage value and the capacitance value are obtained by decomposing in a scientific counting method.

In the steps S2 and S3, alternative product selection is performed according to the rule of substitutability of each parameter.

In step S4, the formula for calculating the product score is:

S_i＝Q_i+M_i+P_i；

wherein S is_iIndicates the product score, Q, of the ith product_iRepresents the purchase quantity score, M, of the ith product_iIndicates the total purchase price, P, of the ith product_iRepresenting a purchase price point of the ith product; x is the number of_iIndicates the total purchase amount of the ith product, y_iIndicates the purchase total of the ith product, z_iThe purchase price of the ith product is shown, and n is the total amount of the matched products.

In step S5, the specific method for calculating the probability that the user may purchase the product is as follows:

s501, determining classification attributes of the naive Bayesian classification as 4 characteristic attributes of product score, purchase quantity score, total purchase price score and purchase price score of the product; the classification category is 2 categories, namely purchase category and non-purchase category; determining test set data and training set data;

s502, training is carried out through a naive Bayes classifier, wherein the input of the classifier is the product characteristics and the purchase condition of a training set, and the output is the purchase condition;

s503, classifying the product feature set in the test set by using a discriminator, and judging whether the product is purchased or not purchased;

s504, sorting according to the purchasing probability of the purchased products and generating a recommendation sequence.

In step S503, the method for determining the product category includes: if P (X | y)₀)P(y₀)＞P(X|y₁)P(y₁) Judging that the product attribute is not purchased; if P (X | y)₀)P(y₀)＜P(X|y₁)P(y₁) Judging the product attribute as purchase;

in step S504, the purchase probability of the product with the attribute of purchase is P (X | y)₁)P(y₁) Wherein X represents a product, y₁Indicates the category is purchase, y₀Indicates the category is not purchased, P (X | y)₀) And P (X | y)₁) Conditional probabilities, P (y), of the categories as unpurchased and purchased, respectively₀) And P (y)₁) Indicating the prior probability of the category being unpurchased and purchased, respectively.

Compared with the prior art, the invention has the following beneficial effects: the invention provides a matching method of MLCC (multilayer ceramic capacitor), which improves the efficiency of capacitance parameter decomposition by firstly extracting brands and original plant models of material description contents input by a user and then decomposing the parameters, improves the precision of capacitance parameter decomposition by decomposing the capacitance parameters of the capacitance description without brands and original plant models according to the naming characteristics of the MLCC, realizes the matching and recommendation of the capacitance parameters, greatly improves the parameter decomposition speed of the MLCC capacitance while keeping higher level of precision, recommends a proper MLCC capacitance for the user through a product purchase possibility calculation structure, and improves the efficiency and precision of MLCC capacitance matching.

Drawings

FIG. 1 is a flowchart illustrating decomposition of MLCC capacitor original factory model parameters according to an embodiment of the present invention;

FIG. 2 is a flow chart illustrating an exemplary MLCC capacitance parameter decomposition according to the present invention;

fig. 3 is a schematic flow chart of extracting capacitance and voltage values by a scientific calculation method in the embodiment of the invention.

Detailed Description

In order to make the technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to specific embodiments and accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention; all other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

11000 real material descriptions in 100 enterprise purchase lists are collected, and the first column in table 1 shows a part of the material descriptions), wherein 10000 material descriptions are related to the MLCC capacitor. Analysis was summarized by a professional hardware engineer: the method comprises a MLCC capacitor class confirmation method process, a brand confirmation method process and an MLCC capacitor key parameter decomposition method process.

TABLE 1 partial MLCC capacitance Material description and parameter decomposition results

In the material description of 10000 MLCC capacitors, there are three cases as follows:

1. original factory type, brand;

2. the original factory type;

3. no original factory model, pure MLCC material description.

Key parameter confirmation of MLCC capacitance: the key parameters of the MLCC capacitor are given by professional hardware engineers, and the specific parameters are as follows:

TABLE 2 key parameter table of MLCC capacitor

	1	2	3	4	5	6	7
								Parameter(s)	Articles and the like	Volume value	Accuracy of measurement	Rated voltage	Package with a metal layer	Material of	Brand

The brand unnecessary decomposition parameter is decomposed in the information given by the user. The category is the parameter which must be decomposed, and the meaning of the decomposition of the description is only after the category is determined. After the class is determined, other key parameters in the description exist, namely decomposition. The input of the user is judged by unifying the spacers in the description, and converting the spacers shown in table 2 and a plurality of continuous space characters into unified characters, namely space characters.

TABLE 3 Interval symbol (part)

	1	2	3	4	5	6	7	8	9	10	……
												Interval symbol	'	,	，	(	)	_	/	\	。	“	……

The capacitance parameter matching of the above three cases is described below with reference to specific embodiments.

First, case 1: the capacitance parameters of the original factory model and the brand are matched;

(1) determining brand

And establishing a brand corresponding relation. Professional hardware engineers and skilled purchasing experts give corresponding relations of various brands, such as:

TABLE 4 Brand alignment relationship Table (parts)

User input

Guogong (huge country)

YAGEO

Yageo

Village field

MuRata

Murate

Three stars

……

Brand

Guogong (huge country)

Village field

Three stars

……

And extracting the brand information input by the user according to the material description input by the user, wherein the extraction sequence is carried out according to the byte length of the brand name input by the user from high to low, and correspondingly outputting the uniform brand name after the brand corresponds. And finally, deleting the extracted brand information input by the user in the byte description input by the original user, thereby avoiding the influence caused by the next parameter decomposition. And (4) performing decomposition by entering a parameter decomposition step of the case 2 without the brand material description.

(2) Determining possible genuine model

And analyzing the characteristics of the MLCC original factory model and the original factory model of other electronic components according to statistics. And primarily screening possible models.

First, segmentation is performed according to a space character based on a description input by a user.

Then, the user description character string is divided according to the spacer, and after the divided substrings are removed from the-, # and # characters, whether the substrings are the combination of pure letters and numbers and whether the length is more than 8 is judged. If yes, the character string is a possible original factory model; otherwise, judging the situation as a situation 3, and performing parameter decomposition on the situation according to a method of a third situation when the situation is not the pure material description of the original plant model.

(3) Determining a series

After the product brand is confirmed, the series names of the corresponding brands are compared according to the MLCC, and after the series names are met, the corresponding original factory model decomposition process is carried out.

(4) Parameter decomposition

The MLCC capacitor prototype decomposition process for TCC series such as three-ring is shown in fig. 1. From the previous steps: the input for this link is a substring x of known brand, possibly genuine model.

Step 1: extracting 1-3 bits of the substring x to determine whether the substring x coincides with "TCC". If yes, returning to the brand, three rings; the classification is as follows: MLCC capacitance. And continuing to enter the parameter decomposition process of the MLCC capacitor. If not, the next step is carried out.

Step 2: and extracting 4-7 bits of the substring x to determine whether the substring x coincides with the English packaging of the MLCC capacitor (the English packaging of the MLCC capacitor is shown in Table 4). If yes, outputting and entering the next parameter decomposition; if not, the key parameters of the original factory model are missing or the original factory model is wrong, and the decomposition is not continued.

TABLE 5 English system packaging table (part) of MLCC capacitor

	1	2	3	4	5	6	7	8	……
										British system package	1005	0201	0402	0603	0805	1206	1812	2512	……

And step 3: the 8-10 bits of the substring x are extracted to determine whether there is coincidence with the material of the MLCC capacitor (the common material of the MLCC capacitor, the error expression that the user may have common errors but is accustomed to, and the corresponding unified output, as shown in Table 6). If yes, outputting the material, and entering the next step; if not, the next step is carried out.

TABLE 6 MLCC materials corresponding table (part)

And 4, step 4: the 11-13 bits of the substring x are extracted, and the 11 th and 13 th bits are judged to be numbers, and the 12 th bit is 'R' or a number. If yes, calculating the capacity value downwards; if not, the next step is carried out. When the 123 bits are ' R ', the capacity value is 11-13 bits of x, the middle 12 bits ' R ' is converted into a decimal point ', and the unit of the corresponding numerical value is pf; when 12 bits are numerical values, the digits 11 to 12 of the volume value x are multiplied by 10 to be a base number, the 13 th bit is a numerical value of a power exponent, and the final result is the volume value in the unit pf.

And 5: the 14 bits of the substring x are extracted whether there is coincidence with the alphabet of the MLCC capacitance representing the precision (the precision alphabet corresponds to the precision, as shown in table 7). If yes, outputting corresponding precision, and entering the next step; and if not, directly entering the next step.

TABLE 7 precision alphabet correspondence table

	1	2	3	4	5	6	7	8	9	10	11	12
													Letters	A	W	B	C	D	F	G	J	K	M	S	Z
Accuracy of measurement	0.05	0.05	0.1	0.25	0.5	1％	2％	5％	10％	20％	20％-50％	20％-80％

The precision output of the MLCC capacitor is unified as shown in the table, and the unit of a decimal part is pf, for example, 0.05 represents +/-0.05 pf; the percentage fraction, e.g. 1%, represents. + -. 1%.

Step 6: and (4) extracting 15-17 bits of the substring x, calculating the rated voltage, wherein the calculation method is the same as the method for calculating the capacitance value in the step 4, and the voltage unit is uniformly V.

The above steps can extract key parameters of MLCC capacitor of three-ring TCC series: class, volume, rated voltage, precision, packaging, material, brand. The steps of decomposing MLCC parameters by original model numbers are approximately the same, except that each brand or series represents the position change of the parameters or the change of substituted letters, and the rest is approximately the same as the steps of decomposing the parameters by MLCC capacitors of a three-ring TCC series.

(5) If the series output is determined to be successful, checking output class parameters, if the parameters are MLCC capacitors, continuously checking parameters of capacitance values, rated voltages, precision, encapsulation and materials, and judging whether all the parameters are successfully extracted, and if the parameters are successfully extracted, returning key parameters and corresponding values of the MLCC capacitors; if the extraction is incomplete, the original user extracts the character string after the brand, and removes the sub-characters after the information extraction is successful to obtain the character string, and the parameter decomposition process of the condition 3 is entered.

Second, case 2: capacitance parameter matching with original factory model

And (3) for the parameter decomposition process of the original factory model of unknown brand, comparing with the parameter decomposition process of the original factory model of known brand, after (3) confirming that the series is different, the brand and the series of the original factory model need to be reversely deduced through the original factory model, and entering into (4) the parameter decomposition process.

The process of reversely deducing the brand and the series of the original factory model through the original factory model is as follows:

first, the corresponding lowest number of brands and series containing the key parameters corresponding to each MLCC capacitor is confirmed. And judging the brand and the series of the substring x which may be the original manufacturer model, and determining a parameter decomposition process. For example, when a certain model TCC0805Y5V105M500DT of the three-ring TCC series is unknown, the basis for judging the brand and the series is as follows:

the length of x is more than 17, 1-3 bits are 'TCC', 4-7 bits belong to English system packaging (table 4), 8-10 bits belong to MLCC capacitor material (table 5), 14 th bit belongs to precision of MLCC capacitor (table 6), and 11-12 th and 15-17 th bits are all pure numbers or a combination of pure numbers plus one 'R'.

If the character string satisfies the above conditions, it can be determined that the sub-character string is a three-ring TCC series MLCC capacitor. For the brand series, professional purchasing and hardware engineer experience, due to the characteristics of the combination method, the overlapped brand series are very few and almost none, and the conditions of the brand and the series can be properly paid and judged for the original factory models, if the length of x is more than 10, 1-3 bits are 'TCC', 4-7 bits belong to English system packaging (table 4) and 8-10 bits belong to MLCC (MLCC capacitor) materials (table 5).

For the brands and series which are difficult to distinguish, the conditions should be strictly card-paired, and professional procurement and hardware engineers are sometimes needed to confirm a large probability of time, such as the series model of the MLCC capacitor of 0603F104Z500CT in Huaxin technology and 0805X104J500NT in Waihouke. The positions of the key parameters are identical: 1-4, packaging; position 5, material quality; 6-8 bits, capacity value; bit 9, precision; 10-12 bits, voltage. Their packaging, capacitance, accuracy, voltage representation, have no distinguishing effect at all.

For example, the material of the Huaxin technology for the material is shown in the following table:

TABLE 8 Material mapping table of Huaxin technology

N

B

D

X

F

S

A

Material of

C0G

X7R

X7E

X5R

Y5V

X6S

X7S

The B, X, F letters correspond to the same letters as those of Fenghua Gaoke. Therefore, the model of a part of original factories belonging to the Huaxin technology can be judged through the rest letters. If the overlap is determined, the 13 th bit is further determined by the experience of professional hardware engineers and professionals, and if the overlap is determined to be C, the 13 th bit is the MLCC of Huaxin technology, because the 15 th bit of the MLCC of Fenghuagaku is C, but the situation is very rare.

After judging the brand and the category, the parameter decomposition is carried out, which is the same as the parameter decomposition process in the case 1.

Third, case 3: and the parameter decomposition of pure MLCC material description without original plant model is realized.

First, the MLCC capacitance class needs to be identified, which includes the following steps:

step 1: there is information such as the name, alias, etc. of the MLCC capacitor, and the MLCC capacitor class name alignment is shown in table 9:

TABLE 9 list of MLCC capacitor class names

1	2	3	4	5	6	7	8	9	……
										MLCC capacitor	Paster capacitor	MLCC	mlcc capacitance	mlcc	CAPCC	Capacitor	STM capacitor	Sheet container	……

According to the accumulation of the names of the MLCC capacitor classes, the alias names of the MLCC capacitor classes can be further expanded. If the name, the alias and the like of the MLCC exist, judging the description as the MLCC, deleting the judged name, the alias and the like, and entering a key parameter decomposition process of the MLCC; if not, the step 2 is entered to continue judging the categories.

Step 2: and judging the MLCC capacitor according to a specific packaging writing method of the MLCC capacitor.

TABLE 10 Special packaging table for MLCC capacitor

1	2	3	4	5	6	7	8	……
									C1005	C0201	C0402	C0603	C0805	C1206	C1812	C2512	……

If the specific package in the table 10 exists, the description can be judged to be the material description of the MLCC capacitor, and the decomposition process of key parameters of the MLCC capacitor is started; if not, the step 3 is entered to continue judging the category.

And step 3: and judging the MLCC capacitor according to the combination of the packaging material and the material. And (4) deducing the MLCC capacitor according to the packaging mode representing the patch and the material representing the ceramic capacitor.

Watch 11 packaging and material watch

	1	2	3	4	5	6	7	8	……
										Package with a metal layer	1005	0201	0402	0603	0805	1206	1812	2512	……
Material of	NP0	C0G	X7R	X5R	Y5V	X7S	X6S	X7T	……

When one of the packages and one of the materials in the table 11 appear in the material description, the material description can be judged to be the material description of the MLCC capacitor, and the decomposition process of key parameters of the MLCC capacitor is started; if not, the step 4 is entered to continue judging the category.

And 4, step 4: unify units of information input by users, requirements of the unified units and the corresponding relation of the system, as shown in table 12:

table 12 unit unified table

μF

uf

nf

Nf

Pf

Kv

KV

v

……

Unit of

pf

v

……

Carry system

1000000

1000

1

1000

1

……

And according to units existing in the material product, establishing unified standard output.

1. Precision: according to the analysis of MLCC capacitance material description, a precision user can be represented by three forms of prefix +/-minus or plus, suffix% and letter. Since the precision of the alphabet representation is the specific attribute of the MLCC capacitor, the description cannot be used as the basis for judging the category when the description is not judged to be the MLCC capacitor. Therefore, the precision before the MLCC is not judged, and a prefix + -, and a suffix% are used as marks. Presence, i.e. indicating that there is precision in the description; absence, i.e. absence of description with respect to precision

2. Packaging: judging whether the package exists or not, judging according to the corresponding relation of the main bodies of the package, and aligning the main bodies of the package according to the following table 13:

table 13 package body correspondence table

And correspondingly inputting corresponding English system packages according to the sequence of MLCC capacitor packages, English system packages and metric system packages. Presence, i.e. encapsulation, in the description; there is no encapsulation in the description.

3. The material is as follows: and judging whether the material exists. The material in the description is judged by the material in table 6. The presence, that is, the material (material for short) of the MLCC capacitor exists in the description; there is no material present, i.e. there is no material in the description.

Whether the capacitor is an MLCC capacitor or not can be judged by a unit or a data parameter type through the following combination.

(1) Capacity pf unit, voltage v unit, precision, material

(2) Two capacity pf units, voltage v units, material

(3) Capacitance pf unit, voltage v unit, precision, package

(4) Two capacitance pf units, voltage v units, package

In the material description, any one of the above conditions exists, and the MLCC capacitor can be judged, and the parameter decomposition of the MLCC capacitor can be carried out. For the description of the MLCC materials which are seriously lost, because the key parameters are incomplete, ambiguity exists in product positioning, namely the product cannot be judged, and measures which are not decomposed are taken for the description of the materials which are not satisfied by the conditions, namely the description of the materials which are not MLCC capacitors or the description of the MLCC capacitors which are seriously lost.

Then, the material description with the judged category being the MLCC capacitor is subjected to parameter decomposition, and the decomposition process is shown in FIG. 2. And (4) describing the uniform symbols and the user MLCC capacitor user materials after class judgment as the input of the stage. The decomposition process comprises the following steps:

step 1: output class, MLCC capacitance.

Step 2: unify the specific precision inputs of the MLCC capacitance that may be present.

TABLE 14 MLCC capacitance special precision corresponding table

+80/20％

-20/+80％

20+80％

+80-20％

+50/20％

-20/+50％

20+50％

+50-20％

……

Accuracy of measurement

+80％/-20％

+50％/-20％

……

And after the correspondence, replacing the special precision description in the character string according to the unified standard.

And step 3: and extracting output materials. Extracting the material in the description according to the material in the table 6, if the material exists, outputting a uniform material symbol, deleting the original symbol which represents the material in the character string, and entering the step 4; and if not, directly entering the step 4.

And 4, step 4: and extracting and outputting the package. Extracting according to the package in table 13, if the package exists, the package in the description is obtained, a unified English system package is output, the original symbol which represents the package in the character string is deleted, and the step 5 is carried out; and if not, directly entering the step 5.

And 5: and extracting the output precision. Extracting the number before the% according to the characteristic of the existence of the precision, if the number exists, judging the number of the numbers, and if the number is 1, directly outputting the precision, such as 10%; if the number is more than 1, the precision range is output from the minimum value to the maximum value, such as 20-80%. If the number of the digits is 1 to less than 1, directly outputting the precision, such as 0.05; when the number is 1 and is more than or equal to 1, directly outputting the precision, such as 10 percent; if the number is more than 2, taking the minimum value, and if the number is less than 1, outputting the precision, such as 0.05; greater than or equal to 1, and output 1%. Proceed to step 6.

Step 6: and unifying the units. According to table 12, the content of the remaining character string is converted into numerical values for the unit portions according to the system, and the representation of different units is unified.

And 7: and extracting the voltage. I.e. the value preceding the extracted voltage symbol v, is present, i.e. is the nominal voltage value in this description. There is no voltage present, i.e. there is no voltage represented in this form in this description.

And 8: and extracting the volume value. That is, the value preceding the capacitance symbol pf is extracted, and the number of values to be determined may appear in the form of capacitance, but the capacitance value of the MLCC capacitor is always larger than the precision value, and therefore the larger value of the values is taken as the capacitance value in the description. There is no value in this description that means there is no value in this form.

And step 9: if the precision is not successfully extracted in step 5 and the number of the values before pf extracted in step 8 is 2, the value with the smaller value is taken as the precision value of the MLCC capacitor.

Step 10: the specific units of the uniform MLCC capacitance (shown in table 15) are shown. In the industrial habit, the unit of the capacitance value in the MLCC capacitor is abbreviated, and if the capacitance value is successfully extracted before step 10, the special units are unified. And corresponding values can be unified into the maximum value in the values after pf, and the capacity value is output.

TABLE 15 Unit of MLCC capacitance

μ

u

n

N

P

……

Unit of

pf

……

Carry system

1000000

1000

1

……

Step 11: if the above steps are not performed, the output of voltage and capacitance is still lacked. The capacitance value and the rated voltage are described in a scientific counting method form in the consideration description. Meanwhile, the precision of the MLCC capacitor is often expressed by letters under the condition that the precision is obtained by experience of professional purchasers and hardware engineers, so that the precision is output on the premise of no precision output. The specific steps are shown in fig. 3, and comprise the following steps:

(1) integer values of length 3 in the descriptive string are extracted.

(2) The letters immediately after the integer are extracted.

(3) If the number of the numerical values is 1. If no capacity value is output, the integer is scientific counting representation of the capacity value; if there is a capacitance value output and there is no voltage output, the integer is a scientific counting representation of the voltage. If the letter exists and is not output with precision at the same time in the letter represented by the precision symbol, outputting the precision represented by the letter.

(4) If the number of the numerical values is 2. Since the precision expression letter C is repeated with the temperature habit expression symbol C and is derived from expert experience, the precision probability expressed with the letter C is extremely low and the temperature probability expressed with the letter C is extremely high, the recognition processing is not performed here with C as the form of precision expression.

After the numerical values of the integers are not provided with letters, the first integer is scientific counting representation of the volume value; the second integer is a scientific counting representation of the voltage. After the integral number, the number of letters is 1, C exists, and the non-C number is processed according to the method that the number of the number is 1. After the numerical value of the integer is reached, the number of the letters is 1, C is absent, and the numerical value of the letters is absent behind the numerical value, so that the integer is represented by scientific counting of the volume value; with a letter and within the precision representation letter, the integer is a scientific counting representation of the voltage. After the numerical value is integrated, the number of the letters is 2, C is not available, the letters are expressed by precision letters, and the corresponding number is represented by scientific counting of voltage; and for other letters, the corresponding number is a scientific counting representation of the volume value.

(5) The digit length is greater than 2. The letters are in the precision representation range, and the corresponding numerical values are scientific counting representation of the voltage; non-alphabetic, corresponding numerical values are scientific counting representations of the numerical values.

(6) And (4) representing the obtained numerical values, and judging a counting method of scientific counting. If the first bit is 0, directly outputting the numerical value represented by the last two bits; if the first digit is not 0, the multiplication of the value represented by the first two digits of the output value by the base number is 10, and the power exponent is the value of the last digit. And all parts having meaning extraction in step 11 are deleted on the basis of the character string input in step 11 as input in step 12.

(7) If there is a capacitance value and no voltage before step 11, and there is no voltage and a capacitance value in step 11, the capacitance value obtained in step 11 is used as the final voltage output.

If there is no capacity value before step 11, and there is a capacity value in step 11, the capacity value in step 11 is output as a capacity value.

If there is no voltage before step 11, and there is voltage in step 11, the voltage in step 11 is outputted as voltage.

If there is no precision before step 11 and there is precision in step 11, the precision in step 11 is output as precision.

Step 12: if the above steps have no precision output, searching the letters of the remaining character strings in the step 11, and if the letters representing the precision exist, outputting the representation of the representation precision of the letters.

Step 13: and in the steps, the parameter name and the corresponding parameter part are obtained and output.

After the product parameters are obtained through the parameter decomposition process, matching can be carried out to obtain corresponding products and alternative products, and then the product purchase possibility is calculated. The method specifically comprises the following steps:

(1) according to the input of the user, the parameters are decomposed, and corresponding products and alternative products are matched.

(2) The corresponding products, i.e. products with all the same parameters, are matched.

(3) And (4) selecting alternative products according to the characteristics of the MLCC capacitor and the alternative rules of various parameters (shown in a table 15).

TABLE 16 alternative selection conditions

1

2

3

4

5

6

7

Parameter(s)

Articles and the like

Volume value

Accuracy of measurement

Rated voltage

Package with a metal layer

Material of

Brand

Relationships between

＝

>＝

<＝

>＝

＝

>＝

(4) And calculating a product score.

And counting the purchasing situations of the users who purchase the corresponding products. The method is characterized in that the method corresponds to the conditions of quantity, price and the like of each material number product purchased by each user and the inventory condition and delivery period condition corresponding to the platform. The method specifically comprises the following steps:

S_i＝Q_i+M_i+P_i； (1)

wherein S is_iIndicates the product score, Q, of the ith product_iRepresents the purchase quantity score, M, of the ith product_iIndicates the total purchase price, P, of the ith product_iRepresenting a purchase price point of the ith product; x is the number of_iIndicates the total purchase amount of the ith product, y_iIndicates the purchase total of the ith product, z_iIndicating the purchase price of the ith product.

(5) Each score of a product is taken as an input of the product, and is expressed as x ═ a₁，a₂，a₃，a₄) The probability that the user is likely to purchase the product is calculated using a naive bayes algorithm.

The first step is as follows: a preparation phase. The classification attributes of the naive Bayes classification are determined as each score of the product, and the total number of the classification attributes is 4. The classification category 2 is purchase and non-purchase. The score characteristics and purchase conditions of 10000 products are used as a training set, and 1000 products and purchase conditions are used as a testing set.

The second step is that: and (5) a classifier training phase. The input is the product characteristics and purchase condition of the training set and the output is the purchase condition.

Calculating the score feature of each description data in the training set based on the feature set and the conditional probability estimation of the purchase condition, wherein 0 represents non-purchase, 1 represents purchase:

P(a₁|y_i)，P(a₂|y_i)，…，P(a_n|y_i) Wherein n is 10000, i belongs to {0, 1 }; (5)

calculating the prior probability P (a)_j|y_i) Means in the category y_iMedium, i.e. purchased or not purchased, product score feature a_jThe probability of occurrence can be solved as:

calculating and counting the occurrence frequency P (y) of each class in the training set sample_i) The solution can be approximated as:

wherein y is_iWhen 0, the sample category is not purchased; y is_iWhen 1, the sample category is purchase.

The third step: and an application stage. And classifying the product feature set in the test set by using a discriminator. The input of the classification method is a discriminator and a feature set of a product to be classified, and the output is the category of the feature set of the product to be classified, namely the category of the product is purchased or not purchased.

Calculating posterior probability P (y) of test set product X_i|X):

Since the sizes of P (X) are fixed and invariable under different classes, only the molecule P (X | y) needs to be compared and calculated when the posterior probability of the skull sample X is calculated_i)P(y_i):

Wherein:

if P (X | y)₀)P(y₀)＞P(X|y₁)P(y₁) Judging that the product attribute is not purchased; if P (X | y)₀)P(y₀)＜P(X|y₁)P(y₁) Judging the product attribute as purchase, and recording the probability P (X | y) at the time of purchase₁)P(y₁). Wherein X represents a product, y₁Indicates the category is purchase, y₀Indicates the category is not purchased, P (X | y)₀) And P (X | y)₁) Conditional probabilities, P (y), of the categories as unpurchased and purchased, respectively₀) And P (y)₁) Indicating the prior probability of the category being unpurchased and purchased, respectively.

(6) And sorting the products with the attributes as purchased according to the probability and carrying out a recommendation sequence.

The invention provides an MLCC capacitor parameter matching method, which adopts a key parameter decomposition method, combines the industrial experience of a plurality of professionally purchased years and the professional knowledge of hardware engineers, thereby summarizing a set of method suitable for MLCC capacitor decomposition.

(1) By decomposing 11000 pieces of data, parameter decomposition is performed. The accuracy of the decomposition results is shown in the following table:

TABLE 17 table of results of parameter decomposition experiments

By the method, the key parameters of the MLCC capacitor are decomposed, and the accuracy is high.

(2) The common time for decomposing the parameters of 11000 data is 18.07s, about 1 min/data is needed in a manual mode, and 183.3 hours are needed for decomposing 11000 data under the condition that fatigue equivalent rate is reduced because time is normal. Therefore, the efficiency of the method is improved by 600 times compared with that of the manual method. And the method is not influenced by working experience and working time.

(3) The accuracy of the product purchase prediction is more than 93%. The purchasable products, the alternative purchasable products and the like can be effectively and quickly recommended to the user, and the working efficiency and the user experience are improved.

In summary, the present invention provides a matching method for MLCC capacitance parameters, which decomposes material description content input by a user to realize matching and recommendation of capacitance parameters, greatly increases the parameter decomposition speed of MLCC capacitance while maintaining a high level of accuracy, recommends a suitable MLCC capacitance for the user through a product purchase probability calculation structure, and increases the matching efficiency and accuracy of MLCC capacitance.

The embodiments of the present invention have been described in detail with reference to the accompanying drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the gist of the present invention.

Claims

1. A method for matching MLCC (multilayer ceramic capacitor) capacitance parameters is characterized by comprising the following steps:

2. The matching method for the MLCC capacitance parameters according to claim 1, wherein in the step S1, the specific method for judging whether the material description has a brand and a genuine manufacturer model is as follows:

3. The method for matching MLCC capacitance parameters according to claim 1, wherein in step S1, the specific method for determining whether the MLCC capacitance is the MLCC capacitance according to the material description input by the user is as follows:

4. The MLCC capacitance parameter matching method according to claim 1, wherein the step S2 specifically comprises the steps of:

5. The MLCC capacitor parameter matching method according to claim 4, wherein the English system package character bits in the sub-character string corresponding to the original factory model number are 4-7 bits, the material character bits are 8-10 bits, the capacitance character bits are 11-13 bits, the precision character bits are 14 bits, and the rated voltage character bits are 15-17 bits.

6. The MLCC capacitance parameter matching method according to claim 1, wherein the step S3 specifically comprises the steps of:

7. The method as claimed in claim 6, wherein in step S304, if the precision is not successfully extracted and the character string representing the capacitance value in the material description is extracted in step S307, the number of the numerical values before pf is 2 and the corresponding numerical values are all less than or equal to 10, the value with the smaller numerical value is used as the precision value of the MLCC capacitor;

8. The method as claimed in claim 1, wherein in steps S2 and S3, the alternative product selection is performed according to the alternative rule of each parameter.

9. The matching method for the MLCC capacitance parameters according to claim 1, wherein in step S4, the formula for calculating the product score is:

S_i＝Q_i+M_i+P_i；

10. The matching method for the MLCC capacitance parameters of claim 1, wherein in step S5, the specific method for calculating the probability that the user may purchase the product is as follows:

s504, sorting according to the attribute, the purchasing probability of the purchased products, and generating a recommendation sequence;