CN113077107A - Electric power material configuration demand prediction system - Google Patents
Electric power material configuration demand prediction system Download PDFInfo
- Publication number
- CN113077107A CN113077107A CN202110418092.5A CN202110418092A CN113077107A CN 113077107 A CN113077107 A CN 113077107A CN 202110418092 A CN202110418092 A CN 202110418092A CN 113077107 A CN113077107 A CN 113077107A
- Authority
- CN
- China
- Prior art keywords
- module
- data
- packet
- calculating
- label
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
Abstract
The invention relates to the field of information computing information systems, in particular to a power material configuration demand prediction system. The invention is realized by the following technical scheme: a power material configuration demand prediction system comprises a reverse calculation module for generating a reverse prediction result, and the data prediction system further comprises: the data cleaning module is used for cleaning the basic data; the forward calculation module is used for calculating and generating a forward prediction result; the weight adjusting module is used for adjusting the weight coefficient; a result generation module for generating a final prediction result based on the forward prediction result, the backward prediction result and the weight coefficient; the invention aims to provide a power material configuration demand prediction system which not only carries out multi-channel processing on basic data in a database and improves the effectiveness and pertinence of data use, but also carries out a more personalized calculation mode on the data, so that the prediction result is accurate and the referential property is high.
Description
Technical Field
The invention relates to the field of information computing information systems, in particular to a power material configuration demand prediction system.
Background
The electric power system is mainly responsible for the operation and construction of electric power facilities, and in the operation and construction process, it is the problem that electric power system operation main part was concerned about to reduce the operation cost, improve economic efficiency, promote goods and materials availability factor.
Before the construction of electric power facilities, materials of a project of a distribution network need to be prepared, the requirement of the distribution network materials is often predicted by the full time of a material plan, the materials with strong relevance to the project form a final purchasing requirement plan result after being cooperatively confirmed with a city bureau, and the process is only predicated by the experience of people, so that the problem that the execution period of a frame protocol is long or the material is broken is easily caused.
Therefore, for example, chinese patent document No. CN103903070B discloses an application system resource demand measuring and calculating system, which can use a big data platform to perform intelligent data collection and intelligent analysis on the material demand of a project, and predict the material demand of the project. The demand measuring and calculating system usually comprises a data acquisition system, a data calculation system and a measuring and calculating result issuing system. The data calculation system calculates the data by using a big data intelligent algorithm preset by software engineering technicians and generates a final prediction result.
On the one hand, however, distribution network projects are various and change rapidly, and the basic input data related to the distribution network projects and used as prediction results are various, large in number and complex in situation. If the amount of data is large, the correctness and applicability of the data will be affected, and correspondingly, the calculated material requirement result will be reduced in accuracy. On the other hand, the situation of the distribution network project is complex, the existing computing system is often single in algorithm, the same algorithm module is adopted for different types of data, and the flexibility and the adaptability are not high.
Disclosure of Invention
The invention aims to provide a power material configuration demand prediction system which not only carries out multi-channel processing on basic data in a database and improves the effectiveness and pertinence of data use, but also carries out a more personalized calculation mode on the data, so that the prediction result is accurate and the referential property is high.
The invention is realized by the following technical scheme: a power material configuration demand prediction system comprises a reverse calculation module for generating a reverse prediction result, and the data prediction system further comprises:
the data cleaning module is used for cleaning the basic data;
the forward calculation module is used for calculating and generating a forward prediction result;
the weight adjusting module is used for adjusting the weight coefficient;
a result generation module for generating a final prediction result based on the forward prediction result, the backward prediction result and the weight coefficient;
the forward direction calculation module includes a forward direction calculation module,
the total input module is used for inputting the investment total of the project main body;
the sample period data calling module is used for calling the standard packet data in the sample period in the database;
the sample period packet marking ratio calculating module is used for calculating a packet marking ratio coefficient in the packet marking data in the sample period;
the forward result generating module is used for calculating and generating the forward result according to the bid package proportion coefficient and the project main body investment total amount;
the data cleansing module comprises a data cleansing module and a data cleansing module,
the punching and canceling processing module comprises a punching and canceling identification module and a punching and canceling marking module which adds an identification label to punching and canceling data in the material consumption database;
the abnormal value processing module comprises an abnormal value searching module and an editing module for correcting the abnormal value;
and the negative value removing processing module comprises a stage negative value identification module for searching for the consumption in a preset stage as a negative number.
Preferably, the search range of the abnormal value identification module for the data includes the offset data and the non-offset data.
Preferably, the abnormal value identification module comprises an out-of-threshold identification module, and the out-of-threshold identification module is used for identifying data of the material consumption exceeding a consumption threshold preset by the system.
Preferably, the negative value removing processing module includes an adjacent stage integrating module for automatically integrating the stage data identified as the negative value with the adjacent stage.
Preferably, the data cleaning module further comprises a big data processing module, and the big data processing module comprises a custom editing module for supporting a user to custom identify the big data.
Preferably, the bidding packet data called by the sample period data calling module includes amount data corresponding to the bidding packet, the investment total amount of the project main body input by the total amount input module is the investment amount total amount, and the bidding packet proportion coefficient calculated by the sample period bidding packet proportion calculating module is the bidding packet amount data/investment amount total amount.
Preferably, the weight calculation module includes:
the weight mode editing module is used for inputting two or more weight coefficient values in a self-defined mode;
the test result generation module is used for generating different test results according to different weight coefficient values;
the comparison and sorting module is used for comparing the test result with the real result and sorting the test result from high to low according to the data similarity;
and the weight mode determining module is used for determining the weight coefficient corresponding to the test result with the highest similarity of the sorted data as the final weight coefficient.
Preferably, the reverse calculation module includes a characteristic tag adding module for adding a characteristic tag to the packet and a characteristic matching module for matching a corresponding algorithm based on data of the characteristic tag.
Preferably, the characteristic tag adding module comprises a volatility calculating module and a continuity calculating module, the volatility calculating module is used for adding a volatility tag for reflecting the stability of the tag packet to the material, and the continuity calculating module is used for adding a continuity tag for reflecting the continuity condition that the data of the tag packet is not zero to the material.
Preferably, the continuity calculation module includes an interval index calculation module, the interval index calculation module is configured to sum all interval additional numbers in the sample period of the material, the interval additional number is an interval month number of a non-zero standard packet data before the non-zero standard packet data, the continuity calculation module further includes a warehouse-out frequency calculation module, the warehouse-out frequency calculation module is configured to identify a number of the standard packet data of the material in the sample period, which is not zero, and if the number is smaller than a preset threshold, the continuity label is directly defined as none.
Preferably, the inverse calculation module comprises an algorithm module, and the algorithm in the algorithm module is an ARIMA algorithm, a TBATS algorithm or a Prophet algorithm.
Preferably, the volatility calculating module comprises a variation coefficient calculating module, the variation coefficient is calculated according to an average value and a standard deviation of all the standard packet data in the sample period, the volatility calculating module further comprises a blank deleting module, the blank deleting module is used for identifying the number of the standard packet data of the material in the sample period, which is not zero, and if the number is smaller than a preset threshold value, the volatility label is directly defined as none.
Preferably, the characteristic tag adding module further comprises a seasonal calculating module, and the seasonal calculating module comprises a monthly change calculating module for calculating a maximum value of a standard monthly coefficient in a year, wherein the monthly coefficient is an average value of the month in a sample period divided by all average values in the sample period.
Preferably, the characteristic label adding module further comprises an importance calculating module, and the importance calculating module comprises an amount sorting module for sorting sample period amount ratio values, wherein the sample period amount ratio values are obtained by dividing all shipment amounts of the goods in the sample period by the total shipment amounts of all the goods in the sample period.
In summary, the invention has the following beneficial effects:
1. the data automatically synchronizes the service data of T +1 through the timing script, and the operation efficiency is high.
2. The data of the database is cleaned by using the four modules, namely the offset processing module, the abnormal value processing module, the negative value removing processing module and the important data processing module, so that the objectivity and the accuracy of the data are ensured, and a foundation is provided for material demand calculation of a subsequent measuring and calculating system.
3. The punching and canceling processing module can automatically recognize positive and negative punching and canceling data in the database.
4. The abnormal value processing module can judge the error data in the database and correct the error data.
5. The important data processing module supports an operation user to mark out stage important engineering data with contingency, so that objectivity of material demand calculation is improved.
6. A forward algorithm and a reverse algorithm of top-down logic and bottom-up logic are adopted for the prediction result, so that the accuracy of the prediction result is high, and the reliability and the practicability of the prediction system are greatly improved.
7. And different weight mode calculation comparison modes are adopted for the forward prediction result and the backward prediction result to obtain the optimal weight calculation mode, so that the final prediction result is more accurate.
8. And proper characteristic analysis is carried out on different materials, and different algorithm processing in the later period is matched according to the characteristics of the materials, so that the effectiveness and pertinence of the later-period prediction result are improved, and the accuracy of the prediction system is improved.
9. The characteristic is processed by adding a label value, so that the system identification and the system matching are facilitated.
10. The characteristic label contains comprehensive consideration of volatility, seasonality, continuity and importance, so that the algorithm matching is more targeted.
Drawings
FIG. 1 is a general block diagram according to a first embodiment;
FIG. 2 is a schematic diagram of a data cleansing module;
FIG. 3 is a schematic diagram of an outlier identification module;
FIG. 4 is an architecture diagram of a forward compute module;
FIG. 5 is a block diagram of a reverse computing module;
FIG. 6 is a block diagram of a property tag add module;
FIG. 7 is a schematic diagram of the weight calculation module.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings.
The present embodiment is only for explaining the present invention, and it is not limited to the present invention, and those skilled in the art can make modifications of the present embodiment without inventive contribution as needed after reading the present specification, but all of them are protected by patent law within the scope of the claims of the present invention.
Embodiment 1, a power material allocation demand prediction system, as shown in fig. 1, includes five modules, which are a data cleaning module, a forward calculation module, a reverse calculation module, a weight calculation module, and a result generation module. The functional modules referred to herein can be implemented by conventional code writing, and those skilled in the art can write them by using C language, JAVA language in the prior art.
The data calculation module is used for screening, filtering, cleaning and other actions on the data in the original database before formal calculation, so that the objectivity and accuracy of the data are guaranteed, and a good data basis is provided for subsequent demand prediction calculation. The forward calculation module and the backward calculation module respectively adopt a top-down calculation mode and a bottom-up calculation mode, and the forward prediction result and the backward prediction result are calculated and generated by the two modules through two completely different logic modes. The forward prediction result and the backward prediction result are not final results reflecting the power material allocation requirements in the present case, but need to be adjusted by using the weight calculation module. The weight calculation module is used for determining a reasonable and most appropriate weight coefficient, namely, the weight coefficient is used for adjusting the weight proportion of the forward calculation module and the reverse calculation module, and finally, the result generation module generates a final result of the final power material configuration requirement through the forward prediction result, the reverse prediction result and the weight coefficient.
Each of these modules is described in further detail below.
As shown in fig. 2, fig. 2 is a data cleansing module. The data import module is used for periodically importing data related to material consumption in a total data table in the data center into a material consumption database. Because the data volume is huge, a better method is adopted, a real-time update mode is not adopted, and the T +1 business data is automatically synchronized through a timing script, so that the import flow is optimized.
In the power system, materials are often cables, electric cabinets, switch cabinets and other objects, the historical data are stored in a material consumption database in a project form and a time form and serve as a data source and a data base for demand measurement and calculation of the demand measurement and calculation system, and in the technical scheme, the data are mainly cleaned, so that the correctness and the applicability of the data are improved.
And importing the material consumption data into a material consumption database, scanning and identifying all data by a punching and canceling identification module in the punching and canceling processing module, and judging whether forward and reverse punching and canceling data exist. For example, when an electric cabinet is taken out of 200 units in a month and returned to 200 units in the next month, the two pieces of data form forward and backward offset data, and the offset marking module automatically marks the two pieces of information for a system user to check. The identification and search function is realized by software, and a specific code writing mode is the prior art of practitioners in the software field, which is not limited and described herein. The cancellation marking module only marks the cancellation data, and specific operations after marking, such as deletion, combination, resetting and the like, are different in operation mode in different material projects, and an operator autonomously selects autonomous operation according to a specific application scene.
The outlier processing module, unlike the offset processing module, identifies and processes outliers in the table. The abnormal values are formed by many reasons, such as human input errors, data system storage errors and consumption data in an abnormal stage, and the existence of the data also brings errors for the subsequent data measurement. As shown in fig. 2 and 3, in the present application, the abnormal value processing module includes an abnormal value identification module and an editing module. The outlier identification module may include a single bit error identification module and an out-of-threshold identification module. The two methods are different, and the former method is often caused by the error of the input data due to the error of the data unit by the input personnel. For example, 3 km, unit is km, non-meter, input data of input personnel is 3000, and data error is caused. The out-of-threshold identification module presets a normal and reasonable data range, such as 2-8 km, for the system operator, and the data exceeding the value range is often not in accordance with the actual real situation, so that the possibility of data error exists. The two search functions are also realized by software programming, and the specific programming mode is the content of the prior art, which is not described herein again.
The abnormal value processing module comprises an editing module which supports a user to automatically or manually correct the identified abnormal value. The data traversed by the abnormal value processing module is all data, namely whether the data is the offset data identified by the offset processing module or not, the identification traversal of the abnormal value is carried out.
The negativity value removing processing module also performs traversal search on the data, and it should be noted that the traversal of negativity values is in units of "stages". For example, the material consumption within a week or within a month, the material consumption should not be negative, and during the data system entry, there may be negative values in the database for various reasons. For example, when the user leaves the warehouse 20 at month 4 and returns to 10 at month 5, the system entry may enter data in a format of 20 at month 4 and 10-5 at month 5. After the negative value is identified, the data can be processed in various ways, such as merging and integrating the data by an adjacent stage integrating module, and smoothing. Specifically, the module performs combination calculation on the data of 4 months and 5 months, and after combination, the two pieces of data are modified to be 10 in 4 months and 0 in 5 months.
The important data processing module comprises a custom editing module, which means that a certain known stage important project appears in a history stage, due to the existence of the project, the material consumption of the stage is extremely large, and the data has contingency and can influence the objectivity of a prediction result. In the scheme, an operation user uses a custom editing module to label and identify the data corresponding to the major projects, and the data is not limited to be subjected to smoothing processing or deleting during operation.
Thus, after all data is cleaned, the processed data is used as a usable, effective and historical database to provide a data source for the subsequent calculation of the forward calculation module and the backward calculation module. The data used by the forward calculation module and the backward calculation module are from the historical database processed by the data cleaning module.
The calculation modes of the forward calculation module and the backward calculation module are further elaborated below.
As shown in fig. 4, fig. 4 is a forward computing module, which adopts a top-down logic computing manner and includes four modules. The total amount input module needs a user to input the total amount in a customized manner, and the total amount may be a total amount of regular investment of business units, and in this embodiment, may be a total planned investment of 2021 year in whole province of Zhejiang province, for example, 10 hundred million. The sample period data calling module obtains the tagged data from the historical database.
In the scheme, the required materials are materials, and in the field of electric power, the common materials are components such as a 10KV transformer (dry type), a distribution box, a 10KV ring net cage, an alternating current wall bushing and the like. The standard data is-the quantity of goods and materials out of the warehouse or the quantity of actual use in the statistical period. In the present case, the statistical period is defined as one month. And the logic main body object of the goods and materials or the label bag is defined to the level of a 10KV transformer (dry type), a distribution box, a 10KV ring net cage, an alternating current wall bushing and the like. Taking a 10KV transformer (dry type) as an example, the upper layer of logic main body is a transformer (dry type), and the lower layer of logic main body is a 10KV transformer (dry type) 400A, a 10KV transformer (dry type) 200A, a 10KV transformer (dry type) 300A, and the like. In this embodiment, defining the logical subject object of the material or the label package at this layer is a comprehensive consideration of the applicant based on the data running speed, the data sparsity and the characteristic label accuracy during the use and experiment process.
The sample period data calling module obtains the packet data in the sample period from the historical database, wherein the packet data comprises the names of goods and materials, the shipment volume/use volume in each month and corresponding money data. The sample period may be naturally defined, and in this case, the sample period is defined as the first three years of the prediction period, that is, data of 36 months in total of 2018, 2019 and 2020. And then, the sample period packet proportion calculation module counts the proportion of the money of each material in the sample period, for example, the proportion data of the distribution box is calculated in a way that the total shipment money of the distribution box is divided by the total shipment money of all materials in the 36 months.
Then, the forward result generating module directly multiplies the total investment amount of the user in 2021 years input by the total amount input module by the total amount data of each material calculated by the sample period packet ratio calculating module to obtain a certain material, for example, the total required amount of the distribution box in 2021 years, or the required amount of each month in 2021 years, or directly obtain the required amount of the distribution box because the unit price of the distribution box is known.
The prediction data of all materials obtained based on the modules and the method is the forward prediction result in the scheme.
The calculation method of the backward prediction result is complex, and uses a pattern of characteristic analysis and algorithm matching, as shown in fig. 5, the backward calculation module includes four modules, namely a sample period editing module, a data acquisition module, a characteristic tag adding module and a characteristic matching module. The system has the main functions of sorting and adding corresponding characteristic labels to the materials in the power material configuration demand measuring and calculating system, and the system can match corresponding algorithms at the later stage according to the content of the labels.
Through the sample period editing module and the data acquisition module, all standard packages of the materials in the sample period, namely 10KV transformers (dry type), are acquired, the distribution box, the 10KV ring net cage and the alternating current wall bushing acquire 36 corresponding standard packages, and the content of the 36 standard packages is the actual delivery quantity of the materials in each month in the 3 years. And the characteristic label adding module is used for adding a corresponding characteristic label to the material after analyzing and calculating the 36 samples of the material.
In this embodiment, four characteristic tag adding modules are adopted, which are respectively a volatility calculating module, a seasonal calculating module, a continuity calculating module and an importance calculating module, and are respectively used for adding a volatility tag, a seasonal tag, a continuity tag and an importance tag to the material.
The property tag addition module is described in detail below.
As shown in fig. 6, the volatility calculating module includes a coefficient of variation technique module and a blank reduction module. The blank deletion module is used for searching and confirming whether the 36 standard packets of the material (such as the distribution box) have 0 values, wherein 0 means that the distribution box does not have the shipment in the month. The blank deleting module calculates the number of non-0 values in the 36 standard packets, and judges whether the number is smaller than a preset threshold value. For example, the preset threshold value of the material of the distribution box is 2, the 0 value in 36 standard packets is 35, only 1 non-0 value is needed, and at this time, the fluctuation calculation is not necessary, and the fluctuation label is directly defined as 'none'.
If 0 in 36 standard packages of the distribution box is only 10, and 26 standard packages are non-0 values, the variation coefficient calculation module is needed to calculate the variation coefficient of the material of the distribution box. The variance may be the standard deviation of the 36 values, and then divided by the average of the 36 values, and according to the variance, the user may divide the volatility into several intervals and match different tag data, for example, divide the volatility into 5 steps, which are low, medium, high, and high, respectively. The low, medium, high and high are the volatility characteristic label data of the material. So far the computation and addition of the volatility label is completed.
The seasonal calculation module comprises a monthly change calculation module for calculating the maximum value of the monthly coefficient. The monthly coefficient is calculated by dividing the average shipment for the month over 3 years by the average shipment for all months over 3 years. For example, the monthly coefficient for the 3 months of the distribution box: the shipment amounts in 3 months of the 3 years 2017, 2018 and 2019 are 300, 400 and 500 respectively, and the average number in 3 months is 400. And in the distribution box, in the three years, the average of monthly goods output of 36 months is 320, and the monthly coefficient of 3 months is 1.25. And similarly, the monthly coefficients of other months can be obtained, namely the 12 monthly coefficients of the distribution box are obtained together, and then the maximum value is taken, and the maximum value is the final result value required by the monthly change calculation module. Then, depending on the value of the result, the user can divide the seasonality into several intervals and match different tag data, e.g. divide the seasonality into 3 bins, low, medium, high, respectively. The low, medium and high are seasonal characteristic label data of the material. The calculation and addition of seasonal tags is now complete.
The continuity calculation module comprises an ex-warehouse frequency calculation module and a separation index calculation module. The warehouse-out frequency calculation module is similar to the blank deletion module in function and is used for calculating the times that the shipment quantity is not 0. When the number of months for which the shipment is not 0 is less than a threshold, for example less than 4, then the continuity label is defined as "none".
And when the quantity of the months with the shipment quantity not being 0 is larger than the threshold, the interval index calculation module is used for summing all the interval additional numbers in the sample period of the material, and the interval additional number is the interval month number of the non-zero standard packet data before the non-zero standard packet data. Distance shows that for the example of distribution box shipment data of 1-6 months, if 200, 300, 200 are respectively sent out in 1-6 months, the 6 months have no non-zero data, i.e. no additional number is separated. If there are 200, 0, 300, 0, 200 for months 1-6, respectively, where months 2 and 4 are both 0, there will be an additional number 1 apart for months 3 and 5, in this example, the additional number 2 apart for the 6 months. (for the definition of patent documents, the understanding of examiners is convenient, the description is made specifically, and the principle of the scheme is the same as that of the scheme that the interval of the party is 1, the interval of the party is not 2 or more than 2, and the 36 numbers have 35 basic quantities, but the understanding of the examiners is more convenient)
Similarly, depending on the size of the value that is separated by the sum of the additional numbers, the user may divide the continuity into intervals and match different tag data, for example, divide the continuity into 5 steps, low, medium, high, respectively. The low, medium, high and high are the continuous characteristic label data of the material. By this time the computation and addition of the continuity label is complete.
The importance calculating module comprises a bid package amount sorting module which is used for calculating the importance of the amount of the goods and materials. The specific method is to calculate the sum of all the outgoing money of the material (distribution box) in a sample period, namely 36 months, then calculate the sum of all the outgoing money of all the material in the sample period, namely 36 months, and then divide the former by the latter to obtain the result of the ratio of the money of the material. And the money amount sorting module sorts the money amount ratio results of all the materials according to the numerical value. Then, according to the value of the result, the user can divide the importance into several intervals and match different tag data, for example, divide the importance into 3 grades, low, medium, and high, respectively. The low, medium and high are the importance characteristic label data of the material. By this time the calculation and addition of the importance labels has been completed.
In the above, the characteristic tag adding module has added the corresponding numerical values to the four characteristic tags of each material. The characteristic matching module comprises a mapping editing module and an algorithm module, wherein the algorithm module is an intelligent algorithm used for calculating data later, the mapping editing module is a mapping relation between different label data and different algorithms, and the mapping relation and the intelligent algorithm are edited by a user in a self-defining way.
For example:
if seasonal = medium or high; then model selection = linear and non-linear seasonal combinatorial algorithm models.
If continuity = none; volatility = medium or high; then model selection = discontinuous and seasonal combined algorithm model.
If seasonal = none; volatility = medium; importance = low; the then model selects an autoregressive combined algorithm model = trending and seasonality.
It should be noted that, in the present application, each functional module is implemented by software, and a specific code writing manner is the prior art of practitioners in the software field, which is not limited and described herein.
Characteristic analysis and algorithm matching are completed, algorithms such as a linear and nonlinear seasonal combined algorithm model, a trend and seasonal autoregressive combined algorithm model, a discontinuous and seasonal combined algorithm model, an ARIMA algorithm model, a TBATS algorithm model and a Prophet algorithm model are all the prior art, and the required quantity of materials can be obtained according to the standard packet data content and the corresponding matched algorithm.
The prediction data of all materials obtained based on the modules and the method is the reverse prediction result in the scheme.
At this point, both the forward prediction result and the backward prediction result are calculated. At this time, a weight calculation module is required to determine the weight coefficient. As shown in fig. 7, the weight calculation module also includes four modules. The weight mode editing module is used for a user to define a plurality of different weight distribution modes. For example, in the present embodiment, the user defines five patterns, i.e., a forward predictor weight and a reverse predictor weight as (100%, 0%), (70%, 30%), (50% ), (30%, 70%), (0%, 100%), respectively. According to the above, the system can calculate the forward prediction result and the backward prediction result corresponding to 2021 years according to the data of 2018, 2019 and 2020. Here, the system can calculate the forward prediction result and the backward prediction result corresponding to 2019 according to data of 2016, 2017 and 2018. At this time, the test result generation module generates five different final prediction results in 2019 according to the forward prediction result and the backward prediction result in 2019 and five weight distribution modes. The data required in 2019 is known data and already exists in a historical database, and at the moment, the comparison and sorting module compares the 2019 final prediction results of the five weight distribution modes with the 2019 real historical data and sorts the results from high to low according to the matching degree. For example, in the present embodiment, the pattern with the highest matching degree is (30%, 70%), the matching degree reaches 86%, and the weight pattern determination module determines (30%, 70%) as the final weight pattern.
By now, the three data of the forward prediction result in 2021, the backward prediction result in 2021 and the optimal weight mode are determined, and the result generation module generates the final data of the supplies demand in 2021 based on the three data.
Claims (10)
1. A power material allocation demand prediction system comprises a reverse calculation module for generating a reverse prediction result, and is characterized in that the data prediction system further comprises: the data cleaning module is used for cleaning the basic data; the forward calculation module is used for calculating and generating a forward prediction result; the weight adjusting module is used for adjusting the weight coefficient; a result generation module for generating a final prediction result based on the forward prediction result, the backward prediction result and the weight coefficient; the forward calculation module comprises a total amount input module for inputting the investment total amount of the project main body; the sample period data calling module is used for calling the standard packet data in the sample period in the database; the sample period packet marking ratio calculating module is used for calculating a packet marking ratio coefficient in the packet marking data in the sample period; the forward result generating module is used for calculating and generating the forward result according to the bid package proportion coefficient and the project main body investment total amount; the data cleaning module comprises a punching and canceling processing module, and the punching and canceling processing module comprises a punching and canceling identification module and a punching and canceling marking module for adding an identification label to punching and canceling data in the material consumption database; the abnormal value processing module comprises an abnormal value searching module and an editing module for correcting the abnormal value; and the negative value removing processing module comprises a stage negative value identification module for searching for the consumption in a preset stage as a negative number.
2. The system for cleaning data measured according to the demand for power material allocation of claim 1, wherein: the abnormal value identification module searches the data in a range including impulse data and non-impulse data.
3. The system for cleaning data measured according to the demand for power material allocation of claim 1, wherein: the abnormal value identification module comprises an out-of-threshold identification module which is used for identifying data of the resource consumption amount exceeding a consumption threshold preset by the system.
4. The system for cleaning data measured according to the demand for power material allocation of claim 1, wherein: the negative value removing processing module comprises an adjacent stage integration module used for automatically integrating the stage data identified as the negative value with the adjacent stage.
5. The system for cleaning data measured according to the demand for power material allocation of claim 1, wherein: the data cleaning module also comprises a big data processing module, and the big data processing module comprises a custom editing module for supporting a user to custom identify the big data.
6. The system according to claim 1, wherein the system comprises: the bidding packet data called by the sample period data calling module comprises amount data corresponding to the bidding packets, the project main body investment total amount input by the total amount input module is the investment amount total amount, and the bidding packet proportion coefficient calculated by the sample period bidding packet proportion calculating module is the bidding packet amount data/investment amount total amount.
7. The system of claim 1, wherein the weight calculation module comprises: the weight mode editing module is used for inputting two or more weight coefficient values in a self-defined mode; the test result generation module is used for generating different test results according to different weight coefficient values; the comparison and sorting module is used for comparing the test result with the real result and sorting the test result from high to low according to the data similarity; and the weight mode determining module is used for determining the weight coefficient corresponding to the test result with the highest similarity of the sorted data as the final weight coefficient.
8. The system according to claim 1, wherein the system comprises: the reverse calculation module comprises a characteristic label adding module for adding a characteristic label to the label packet and a characteristic matching module for matching a corresponding algorithm based on data of the characteristic label.
9. The system according to claim 8, wherein the system comprises: the characteristic label adding module comprises a volatility calculating module and a continuity calculating module, the volatility calculating module is used for adding a volatility label for reacting the stability of the label packet to the material, and the continuity calculating module is used for adding a continuity label for reacting the condition that the label packet data is not zero continuity to the material.
10. The system according to claim 9, wherein the system comprises: the continuity calculation module comprises an interval index calculation module, the interval index calculation module is used for summing all interval additional numbers in a sample period of the material, the interval additional numbers are interval month numbers of non-zero standard packet data and previous non-zero standard packet data, the continuity calculation module further comprises a warehouse-out frequency calculation module, the warehouse-out frequency calculation module is used for identifying the number of the standard packet data of the material in the sample period, which is not zero, and if the number is smaller than a preset threshold value, the continuity label is directly defined as none.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110418092.5A CN113077107A (en) | 2021-04-19 | 2021-04-19 | Electric power material configuration demand prediction system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110418092.5A CN113077107A (en) | 2021-04-19 | 2021-04-19 | Electric power material configuration demand prediction system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113077107A true CN113077107A (en) | 2021-07-06 |
Family
ID=76618067
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110418092.5A Pending CN113077107A (en) | 2021-04-19 | 2021-04-19 | Electric power material configuration demand prediction system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113077107A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106779157A (en) * | 2016-11-18 | 2017-05-31 | 中铁第勘察设计院集团有限公司 | The Forecasting Methodology of Regional Energy consumption demand |
| CN107292428A (en) * | 2017-06-07 | 2017-10-24 | 国网浙江省电力公司物资分公司 | A kind of distribution Power Material procurement demand forecasting system |
| CN107292429A (en) * | 2017-06-07 | 2017-10-24 | 上海欧睿供应链管理有限公司 | A kind of Demand Forecast Model system of selection analyzed based on demand characteristics |
| US20190354127A1 (en) * | 2012-07-31 | 2019-11-21 | Causam Enterprises, Inc. | System, method, and data packets for messaging for electric power grid elements over a secure internet protocol network |
| CN111045907A (en) * | 2019-12-12 | 2020-04-21 | 苏州博纳讯动软件有限公司 | System capacity prediction method based on traffic |
-
2021
- 2021-04-19 CN CN202110418092.5A patent/CN113077107A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190354127A1 (en) * | 2012-07-31 | 2019-11-21 | Causam Enterprises, Inc. | System, method, and data packets for messaging for electric power grid elements over a secure internet protocol network |
| CN106779157A (en) * | 2016-11-18 | 2017-05-31 | 中铁第勘察设计院集团有限公司 | The Forecasting Methodology of Regional Energy consumption demand |
| CN107292428A (en) * | 2017-06-07 | 2017-10-24 | 国网浙江省电力公司物资分公司 | A kind of distribution Power Material procurement demand forecasting system |
| CN107292429A (en) * | 2017-06-07 | 2017-10-24 | 上海欧睿供应链管理有限公司 | A kind of Demand Forecast Model system of selection analyzed based on demand characteristics |
| CN111045907A (en) * | 2019-12-12 | 2020-04-21 | 苏州博纳讯动软件有限公司 | System capacity prediction method based on traffic |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Krishna et al. | Sales-forecasting of retail stores using machine learning techniques | |
| CN110231528B (en) | Transformer household variation common knowledge identification method and device based on load characteristic model library | |
| CN112163553B (en) | Material price accounting method, device, storage medium and computer equipment | |
| CN108389069A (en) | Top-tier customer recognition methods based on random forest and logistic regression and device | |
| CN109389270A (en) | Logistics object determination method and device and machine readable medium | |
| CN115547466B (en) | Medical institution registration and review system and method based on big data | |
| CN112860769B (en) | Energy planning data management system | |
| CN108364191A (en) | Top-tier customer Optimum Identification Method and device based on random forest and logistic regression | |
| CN113051291A (en) | Work order information processing method, device, equipment and storage medium | |
| CN109766416A (en) | A kind of new energy policy information abstracting method and system | |
| CN113064992A (en) | Complaint work order structured processing method, device, equipment and storage medium | |
| CN114782065A (en) | Commodity sales volume prediction method and device based on model combination and storage medium | |
| CN116433052A (en) | Intelligent bidding information acquisition data analysis and evaluation system | |
| CN111695979A (en) | Method, device and equipment for analyzing relation between raw material and finished product | |
| CN117196695B (en) | Target product sales data prediction method and device | |
| CN117272995B (en) | Repeated work order recommendation method and device | |
| CN109583773A (en) | A kind of method, system and relevant apparatus that taxpaying credit integral is determining | |
| CN112598227A (en) | Power economic index construction method and system based on power data | |
| CN113077007A (en) | Model matching system for power material configuration demand measurement | |
| CN113077108A (en) | Data prediction system for power material configuration requirements | |
| CN113077107A (en) | Electric power material configuration demand prediction system | |
| CN115081961B (en) | Logistics transport capacity intelligent dispatching method based on big data | |
| CN116912034A (en) | Power distribution network engineering cost data splitting method and auxiliary computing system | |
| CN113298291A (en) | Express delivery quantity prediction method, device, equipment and storage medium | |
| CN114741592A (en) | Product recommendation method, device and medium based on multi-model fusion |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210706 |
|
| RJ01 | Rejection of invention patent application after publication |