Water affair industry medicament purchasing optimization management system based on big data and algorithm

Abstract

The invention discloses a water affair industry medicament purchasing optimization management system based on big data and an algorithm, which belongs to the field of water affair medicament management and comprises a data collection and project archiving module, a medicament sub-item library construction module, a medicament price collection module, a medicament dosage prediction module, a medicament price prediction module and a medicament purchasing optimization management module. Medicament classification is carried out on water affair industries (water supply plants and sewage plants), and the usage amount and the corresponding price of each month in the first three years are counted. The method comprises the following steps: collecting the monthly usage amount of a medicament, obtaining the abnormal degree of the usage amount, analyzing the abnormal usage amount, and constructing a usage amount prediction model to predict the medicament usage amount of each plant station. And constructing a medicament price prediction module based on the medicament price information. And a medicament purchasing optimization management module is constructed by predicting the price according to the medicament predicted usage amount and the plant station margin, so that the purchasing strategy is optimized, and the medicament purchasing management level of the water affair industry is improved.

Description

Technical Field

[0001] This invention relates to the field of water chemical management, and in particular to a water industry chemical procurement optimization management system based on big data and algorithms. Background Technology

[0002] The following problems are currently prevalent in the water industry's chemical procurement management: Data fragmentation: Information such as drug usage, price, and inventory is scattered across different systems, lacking a unified data integration and analysis platform.

[0003] Low prediction accuracy: Traditional usage prediction relies on human experience and does not take into account historical data trends and external factors (such as seasonal changes and water quality fluctuations), resulting in large prediction deviations.

[0004] The procurement strategy is crude: the procurement plan lacks dynamic optimization, often resulting in excessive stockpiling or emergency purchases, which increases inventory costs or delays production.

[0005] Lagging response to price fluctuations: Without a price forecasting model, it is difficult to respond to market fluctuations in a timely manner, resulting in uncontrollable procurement costs.

[0006] In existing technologies, some management systems only implement basic data recording functions and do not deeply integrate big data analysis and algorithm models, thus failing to achieve accurate prediction and dynamic optimization. Therefore, there is an urgent need for an intelligent, data-driven pharmaceutical procurement management system to improve the operational efficiency and economic benefits of the water industry. Summary of the Invention

[0007] The purpose of this invention is to overcome the technical problems existing in the prior art and to provide a water industry chemical procurement optimization management system based on big data and algorithms.

[0008] The objective of this invention is achieved through the following technical solution: A water industry reagent procurement optimization management system based on big data and algorithms includes: The data collection and project archiving module is used to collect and classify historical drug usage data from various water plants. The historical drug usage data includes drug dosage and drug price. The drug category library construction module is used to clean the historical medication data; The reagent price collection module is used to collect and process the prices of reagents purchased by various water plants for different reagents; The chemical dosage prediction module is used to detect anomalies in the monthly chemical dosage of each water plant, and to correct the dosage after anomalies are detected, so as to obtain the chemical dosage data after anomaly treatment; it is also used to build a dosage prediction model and predict the future chemical dosage of each water plant based on historical chemical dosage data and the chemical dosage data after anomaly treatment. The drug price prediction module is used to predict the prices of different drugs; The pharmaceutical procurement optimization management module is used to formulate pharmaceutical procurement plans based on the predicted usage and price of pharmaceuticals.

[0009] In some embodiments, when the data collection and project archiving module categorizes historical medication data, it includes: A hierarchical clustering algorithm is used to classify drugs sequentially based on drug type code, drug subclass, drug description, and technical parameters.

[0010] In some embodiments, cleaning the historical medication data includes: Use MATLAB to write code to clean historical medication data and remove invalid and duplicate data.

[0011] In some embodiments, the abnormal detection of monthly chemical usage at each water plant includes: Calculate the mean and standard deviation of the drug dosage; The threshold for abnormal usage is determined based on statistical rules, and outliers are detected using box plots.

[0012] In some embodiments, the dosage correction after detecting an anomaly includes: Automatic correction: For occasional abnormalities, an algorithm is used to automatically correct them; Manual review: For persistent deviations or complex anomalies, a work order is generated to notify the administrator to confirm the correction plan.

[0013] In some embodiments, the usage prediction model employs the Holt-Winters additive model.

[0014] In some embodiments, the prediction of prices for different pharmaceutical agents includes: Collect drug price data from different sources; Integrate drug price data from different sources to construct time-series characteristics and external influencing factors; Use the following formula to predict prices: in, The predicted price at time t; (This is the intercept item, representing the benchmark price when all factors are zero). The error term represents the impact of other factors not explained by the model on the price; Indicates time as 12-month moving average price: The month-on-month growth rate of the indicator time t-1: Year-on-year growth rate at time t-1: This indicates the volatility of raw material prices. Indicates the normal impact factor. Indicates market factors; The moving average price influence coefficient. The coefficient representing the month-on-month growth rate. The coefficient representing the impact of year-on-year growth rate; This represents the coefficient for the impact of raw material price fluctuations. This represents the policy impact coefficient. This represents the influence coefficient of market factors.

[0015] In some embodiments, the process of developing a drug procurement plan based on predicted usage and predicted price includes: To minimize total procurement costs, a procurement quantity optimization model based on demand coverage is constructed: Objective function: ,in, Indicates the quantity purchased. This represents the predicted price for period t. Constraints: The procurement quantity in each period must meet the current demand. , This represents the demand in period t.

[0016] In some embodiments, a dynamic programming algorithm is used to solve the procurement quantity optimization model.

[0017] It should be further noted that the technical features corresponding to the above embodiments can be combined or substituted with each other to form new technical solutions without conflict.

[0018] Compared with the prior art, the beneficial effects of the present invention are: This invention integrates big data analysis and algorithmic models to detect anomalies in the monthly chemical usage of various water plants. Upon detection of anomalies, usage corrections are performed to obtain anomaly-corrected chemical usage data. A usage prediction model is then constructed to predict future chemical usage at each water plant. Simultaneously, a price prediction model is established to forecast the prices of different chemicals, enabling timely responses to market fluctuations. Finally, based on the predicted usage and prices, a chemical procurement plan is formulated. This achieves an intelligent, data-driven chemical procurement management system, improving operational efficiency and economic benefits in the water industry. Attached Figure Description

[0019] Figure 1This is a flowchart illustrating the workflow of a water industry reagent procurement optimization management system based on big data and algorithms, as described in this invention. Figure 2 This is a schematic diagram of the workflow of the drug dosage prediction module of the present invention; Figure 3 This is a schematic diagram of the workflow of the drug price prediction module of the present invention. Detailed Implementation

[0020] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] It should be noted that the defects in the solutions in the prior art are all the results of the inventors' practice and careful research. Therefore, the discovery process of the above problems and the solutions proposed by the embodiments of this application in the following text should be the inventors' contributions to this application in the process of invention and creation, and should not be understood as technical content known to those skilled in the art.

[0022] In view of the technical problems pointed out in the background art, the present invention provides the following embodiments: Reference Figure 1 In one exemplary embodiment, a water industry reagent procurement optimization management system based on big data and algorithms includes: The data collection and project archiving module is used to collect and classify historical drug usage data from various water plants. The historical drug usage data includes drug dosage and drug price. The drug category library construction module is used to clean the historical medication data; The reagent price collection module is used to collect and process the prices of reagents purchased by various water plants for different reagents; The chemical dosage prediction module is used to detect anomalies in the monthly chemical dosage of each water plant, and to correct the dosage after anomalies are detected, so as to obtain the chemical dosage data after anomaly treatment; it is also used to build a dosage prediction model and predict the future chemical dosage of each water plant based on historical chemical dosage data and the chemical dosage data after anomaly treatment. The drug price prediction module is used to predict the prices of different drugs; The pharmaceutical procurement optimization management module is used to formulate pharmaceutical procurement plans based on the predicted usage and price of pharmaceuticals.

[0023] The implementation methods for the data collection and project archiving module include: ① Collect the dosage and price of chemicals used by each water plant (water supply plant, sewage treatment plant) over the past three years, with the statistical granularity accurate to the month; ② The types of agents are divided into four categories (flocculators, carbon source agents, disinfectants, and others); ③ A hierarchical clustering algorithm is adopted to classify the drugs step by step according to the drug type code, drug subcategory, drug description, and technical parameters. Specifically, the drug code retains a unique identifier for subsequent archiving and association; the drug subcategory, if a rough classification label already exists, can be used as a reference label or auxiliary classification basis; the drug description extracts text keywords (such as usage, ingredients); and the technical parameters select key numerical features such as mass fraction. The first level is divided into four main categories based on drug function (flocculators, carbon source drugs, disinfectants, and others); the second level is further subdivided into drug subcategories based on clustering results (e.g., disinfectants → sodium hypochlorite, chlorine dioxide, ozone, etc.); and the third level is further subdivided according to technical parameters (such as available chlorine content). A three-segment coding system is used, with the format: major category code - minor category code - unique serial number (e.g., XLJ-02-015 represents the 15th drug in the 2nd subcategory of flocculants).

[0024] ④. Assign a regional code to the water industry for different regions; (e.g., using a hierarchical structure of "administrative division code + business type code + facility / region code", for example:) AA: Provincial-level administrative divisions (e.g., Beijing-11) BB: Municipal-level administrative divisions (e.g., Chaoyang District, Beijing - 110105) CC: Business type (e.g., WS-Water, SS-Water Supply, PS-Drainage) Facilities: 001 (Reservoir), 002 (Water Treatment Plant), P01 (Pumping Station) For example: A water supply plant in Chaoyang District, Beijing: 110105-SS-002 The implementation methods for the pharmaceutical sub-library construction module include: ① For different regions and different drugs, the usage of S1 by various manufacturers was cleaned using MATLAB code to remove invalid and duplicate data. (Units were standardized: for example, drug prices were converted to yuan / ton or yuan / kilogram; specifications were matched: drugs of different concentrations or packaging needed to be converted according to the active ingredient). ② Using MATLAB, we cleaned the collected data and removed invalid and duplicate data for the prices of various pharmaceuticals in S1 from different regions and different pharmaceuticals.

[0025] The implementation methods for the drug price aggregation module include: Based on the output of the pharmaceutical category library construction module, the prices of various pharmaceuticals from different manufacturers are aggregated according to different regions and prices in the same period of the previous year. Statistical trend analysis is used to determine prices, and the following definitions are employed: Month-on-month growth rate: (Current period price - Previous period price) / Previous period price; Year-on-year growth rate: (Price in the same period of the same year - Price in the same period of the previous year) / Price in the same period of the previous year; Moving averages (such as 3-period / 6-period smoothing); Linear regression slope: determines the overall trend direction (positive = upward, negative = downward).

[0026] A price trend heatmap is generated using visual charts. The horizontal axis represents the time period, and the vertical axis represents the drug price. The colors are: red (rising), green (falling), and gray (slow).

[0027] Reference Figure 2 The implementation methods of the drug dosage prediction module include: ① Construct a chemical usage database: Group data by plant / station to ensure independent processing of each plant / station's data. Perform statistical analysis on chemical usage at water plants over the past three years (using moving averages (e.g., a 12-month window) to smooth the data and identify potential trends). ② Anomaly Detection Module: Detects anomalies in the monthly usage of chemicals at the plant and obtains the degree of anomaly in the usage. The following are the steps for implementing anomaly detection: a. Calculate statistical indicators: Mean and standard deviation: Calculate the mean and standard deviation of the drug dosage to assess the central tendency and dispersion of the data. Interquartiles: Calculate the first quartile (Q1), median (Q2), and third quartile (Q3), and calculate the interquartile range (IQR = Q3 - Q1).

[0028] b. Determine the threshold. Based on statistical rules, outliers can typically be defined as data points that exceed 1.5 times the IQR range. The outlier range is: Box plot method: This method uses the upper and lower bounds of a box plot to identify outliers. Points that exceed the upper and lower bounds of the box plot are considered outliers.

[0029] Upon detecting an anomaly, the dosage correction module is triggered, and the anomaly type (sudden increase, sudden decrease, or continuous deviation) is recorded.

[0030] ③ Dosage Correction Module logic: Automatic correction: For occasional anomalies (such as a sudden increase / decrease in a single month), an algorithm is used to automatically correct them.

[0031] Manual review: For persistent deviations or complex anomalies (such as anomalies for 3 consecutive months), a work order is generated to notify the administrator to confirm the correction plan.

[0032] Abnormality types and correction strategies: Anomaly types: sudden increase, sudden decrease, persistent deviation. Correction strategies: Sudden increase: moving average (replacing the outlier with the moving average of the previous 3 months); Sudden decrease: median imputation (replacing the outlier with the median of the past 12 months); Persistent deviation: compensation for external factors. .

[0033] Where α, β, and γ are the corresponding weights.

[0034] Database update: Add a new field, Dosage Status (Normal / Corrected / Manually Corrected), to the drug dosage database, retaining both the original and corrected data.

[0035] The implementation methods of the drug dosage prediction module include: Objective: Based on historical data and anomaly-corrected chemical usage data, predict the chemical usage of each plant / station over a future period (e.g., the next 3 months) to provide a scientific basis for procurement planning. This module uses a seasonal triple exponential smoothing model to predict chemical usage based on historical usage data after cleaning, combined with the trend, seasonality, and cyclical characteristics of chemical usage in the water industry. The specific process is as follows: Prediction model construction: The Holt-Winters additive model is used, and the formula is as follows: in, This represents the actual usage in month t (unit: tons / month). This represents the predicted usage value for the next h-month (h=1,2,3 corresponds to the next 3 months). This represents the horizontal component in month t, and the baseline usage. This represents the seasonal component of month t, characterizing periodic fluctuations (such as quarterly or annual variations). Indicates the smoothing parameter. , This indicates the length of the seasonal cycle (s = 12 months).

[0036] Output: Output the predicted annual usage for each plant / station type on a monthly basis.

[0037] Reference Figure 3 The drug price prediction module includes: ① Information collection unit The market price for pharmaceuticals is the lowest value among all prices. Data includes prices from online stores, physical stores, and transaction records. The data also includes the lowest and average prices of current market items, sourced from online stores, physical stores, transaction records, and publicly available market data. (Note: Online stores utilize web scraping software, physical stores use a price inquiry method, and transaction records can be obtained through both web scraping and inquiry.) Details are as follows: Online store data: Use Python web scraping frameworks (such as Scrapy or Selenium) to scrape drug prices from mainstream e-commerce platforms (such as Alibaba and JD.com Industrial Products), extracting fields including: drug name, specifications, unit price (yuan / ton), release time, and supplier information.

[0038] Frequency: Daily scheduled crawling, with data stored in the online_price table of a MySQL database.

[0039] Inquiry data from physical stores: Sign data sharing agreements with suppliers to obtain real-time quotations via API interfaces; or use OCR technology to parse paper quotations and extract price information.

[0040] Frequency: Updated weekly and stored in the offline_price table.

[0041] Publicly available market data: Access the API of commodity trading platforms (such as the Shanghai Chemical Exchange) to obtain price fluctuation data of raw materials (such as liquid chlorine and aluminum sulfate).

[0042] Frequency: Synchronize once per hour and store in the raw_material_price table.

[0043] ② Information processing unit Objective: To integrate multi-source data, construct time-series features and external influencing factors, and provide a high-quality dataset for model training.

[0044] Processing flow: Data fusion: A unified view table is generated by linking online, offline, and raw material price data according to drug codes.

[0045] Feature engineering: Time series characteristics: Extract the moving average (MA), month-on-month growth rate, and year-on-year growth rate for the past 12 months.

[0046] External characteristics: Raw material price volatility (e.g., weekly growth rate of liquid chlorine price).

[0047] Policy factors (such as environmental protection production restriction policies, which are matched with news data by keywords and quantified as 0 / 1).

[0048] Missing value handling: Short-term missing values ​​(≤3 days) were filled using linear interpolation; long-term missing values ​​were replaced using the regional average price of the same drug.

[0049] Data storage: The processed data is stored in a database, partitioned by time, and supports fast querying and backtracking.

[0050] ③Price data prediction module Formula for predicting drug prices: The parameters and calculation formulas are explained below: 1. Core parameters: Predicted price at time t; : Intercept item (base price when all factors are zero); Error term, representing the impact of other factors not explained by the model on price.

[0051] 2. Temporal characteristics: Time is The 12-month moving average price; : Month-on-month growth rate at time t-1 Year-on-year growth rate at time t-1 .

[0052] 3. External influencing factors: Raw material price volatility (e.g., weekly growth rate of liquid chlorine) Normal Influence Factor Market factors (quantitative indicators of supply and demand).

[0053] 4. Coefficient weights: : Moving average price influence coefficient; Impact coefficient of month-on-month growth rate; : Impact coefficient of year-on-year growth rate; : Raw material fluctuation impact coefficient; Policy impact coefficient; : Market factor influence coefficient.

[0054] Output: Adjust the parameter values ​​according to the actual data and output the result.

[0055] The implementation methods for the pharmaceutical procurement optimization management module include: Based on factors such as predicted dosage and price of chemicals, past chemical procurement price trends, and market dynamics, a scientific and reasonable chemical procurement plan is formulated to optimize procurement strategies, reduce procurement costs, improve procurement efficiency, and ensure the stability and economy of chemical supply in the water industry.

[0056] ① Input parameter integration Integrate the following data as the basis for optimization: Forecast usage: Future demand for chemicals at each plant / station (grouped by chemical code and plant / station code).

[0057] Price forecast: Price trends and volatility.

[0058] Data association rules: Associate data from each module by drug code and plant code to build a unified and optimized view.

[0059] ② Dynamic decision-making during the procurement cycle Match the optimal procurement cycle based on price trends: Upward trend: Sign a long-term contract (1-2 years) to lock in the current price and avoid the risk of future price increases.

[0060] If the average annual growth rate is greater than 5%, a two-year contract is recommended; if the growth rate is 3%-5%, a one-year contract is recommended.

[0061] Downward trend: Use short-term procurement (monthly or quarterly) to take advantage of the continuous downward trend in prices and reduce procurement costs.

[0062] The procurement cycle has been shortened to 1-3 months to avoid cost waste due to price declines.

[0063] Stable trend: Adopt a flexible procurement strategy and prioritize the procurement window with the lowest price.

[0064] ③ Procurement quantity optimization model: To minimize total procurement costs, a procurement quantity optimization model based on demand coverage is constructed: Objective function: ,in, Indicates the quantity purchased. This represents the predicted price for period t. Constraints: The procurement quantity in each period must meet the current demand. , This represents the demand in period t.

[0065] Solution: Use dynamic programming algorithm to allocate purchase quantity and optimize by increasing purchase quantity during periods of low prices.

[0066] ④ Dynamic adjustment and feedback mechanism Trend reassessment: Price trends are re-analyzed every quarter to update procurement cycle strategies.

[0067] Abnormal response: If the actual price deviates from the predicted value by more than ±5%, a re-optimization will be triggered.

[0068] If demand suddenly changes (e.g., exceeds the forecast by 20%), a supplementary procurement plan will be generated immediately.

[0069] Results Feedback: Compare actual costs with optimization results monthly to calibrate prediction model parameters.

[0070] ⑤ Output and Execution Procurement plan: Output the procurement cycle, procurement quantity and expected cost for the next 12 months according to drug code and plant code.

[0071] Automated execution: The system automatically generates purchase orders and supports exporting to Excel or PDF formats.

[0072] Risk warning: When price fluctuations exceed expectations or demand is abnormal, an alert will be sent to the management system.

[0073] The above detailed embodiments are a description of the present invention. It should not be considered that the specific embodiments of the present invention are limited to these descriptions. For those skilled in the art, several simple deductions and substitutions can be made without departing from the concept of the present invention, and all of these should be considered to fall within the protection scope of the present invention.

Claims

1. A water industry reagent procurement optimization management system based on big data and algorithms, characterized in that, include: The data collection and project archiving module is used to collect and classify historical drug usage data from various water plants. The historical drug usage data includes drug dosage and drug price. The drug category library construction module is used to clean the historical medication data; The reagent price collection module is used to collect and process the prices of reagents purchased by various water plants for different reagents; The chemical dosage prediction module is used to detect anomalies in the monthly chemical dosage of each water plant, and to correct the dosage after anomalies are detected, so as to obtain the chemical dosage data after anomaly treatment. It is also used to build dosage prediction models and predict future dosages for each water plant based on historical drug usage data and drug usage data after anomaly handling. The drug price prediction module is used to predict the prices of different drugs; The pharmaceutical procurement optimization management module is used to formulate pharmaceutical procurement plans based on predicted usage and predicted prices.

2. The water industry reagent procurement optimization management system based on big data and algorithms according to claim 1, characterized in that, When classifying historical medication data, the data collection and project archiving module includes: A hierarchical clustering algorithm is used to classify drugs sequentially based on drug type code, drug subclass, drug description, and technical parameters.

3. The water industry reagent procurement optimization management system based on big data and algorithms according to claim 1, characterized in that, The cleaning of the historical medication data includes: Use MATLAB to write code to clean historical medication data and remove invalid and duplicate data.

4. The water industry reagent procurement optimization management system based on big data and algorithms according to claim 1, characterized in that, The aforementioned abnormal detection of monthly chemical usage at each water plant includes: Calculate the mean and standard deviation of the drug dosage; The threshold for abnormal usage is determined based on statistical rules, and outliers are detected using box plot methods.

5. The water industry reagent procurement optimization management system based on big data and algorithms according to claim 4, characterized in that, The dosage correction after detecting an abnormality includes: Automatic correction: For occasional abnormalities, an algorithm is used to automatically correct them; Manual review: For persistent deviations or complex anomalies, a work order is generated to notify the administrator to confirm the correction plan.

6. The water industry reagent procurement optimization management system based on big data and algorithms according to claim 1, characterized in that, The usage prediction model adopts the Holt-Winters additive model.

7. The water industry reagent procurement optimization management system based on big data and algorithms according to claim 1, characterized in that, The prediction of prices for different drugs includes: Collect drug price data from different sources; Integrate drug price data from different sources to construct time-series characteristics and external influencing factors; Use the following formula to predict prices: in, The predicted price at time t; (This is the intercept item, representing the benchmark price when all factors are zero). The error term represents the impact of other factors not explained by the model on the price; Indicates time as 12-month moving average price: The month-on-month growth rate of the indicator time t-1: Year-on-year growth rate at time t-1: This indicates the volatility of raw material prices. Indicates the normal impact factor. Indicates market factors; The moving average price influence coefficient. The coefficient representing the month-on-month growth rate. This represents the influence coefficient of the year-on-year growth rate. This represents the coefficient for the impact of raw material price fluctuations. This represents the policy impact coefficient. This represents the influence coefficient of market factors.

8. The water industry reagent procurement optimization management system based on big data and algorithms according to claim 1, characterized in that, The formulation of a drug procurement plan based on predicted drug usage and predicted price includes: To minimize total procurement costs, a procurement quantity optimization model based on demand coverage is constructed: Objective function: ,in, Indicates the quantity purchased. This represents the predicted price for period t. Constraints: The procurement quantity in each period must meet the current demand. , This represents the demand in period t.

9. The water industry reagent procurement optimization management system based on big data and algorithms according to claim 8, characterized in that, The dynamic programming algorithm is used to solve the procurement quantity optimization model.

Images