| บทคัดย่อ |
งานวิจัยนี้ศึกษาวิเคราะห์และพยากรณ์ค่าปริมาณฝุ่น PM2.5 จากการวิเคราะห์ปริมาณฝุ่น PM2.5 ที่เกิดกระบวนการอัตตถดถอยเฉลี่ยเคลื่อนที่ลำดับทั่วไป (General Autoregressive MovingAverage: ARMA(p,q)) เมื่อความคลาดเคลื่อนมีการแจกแจงแบบเลขชี้กำลังของแผนภูมิควบคุมค่าเฉลี่ยเคลื่อนที่ถ่วงน้ำหนักเลขชี้กำลังดัดแปลง (Modified EWMA) และนำเปรียบเทียบกับแผนภูมิควบคุมค่าเฉลี่ยเคลื่อนที่ถ่วงน้ำหนักเลขชี้กำลัง (EWMA)
วิธีการดำเนินการวิจัยนี้แบ่งออกเป็น 2 ขั้นตอน ประกอบด้วย 1) ศึกษาเอกสาร ตำรา และงานวิจัยที่เกี่ยวข้องเพื่อนำไปสู่การพิสูจน์ทฤษฎีเพื่อโดยนำมาวิเคราะห์สูตรสำเร็จ (Explicit formula) ของค่าความยาวรันเฉลี่ย (ARL) จากสมการปริพันธ์ของแผนภูมิควบคุมค่าเฉลี่ยเคลื่อนที่ถ่วงน้ำหนักเลขชี้กำลังดัดแปลง (Modified EWMA) และประมาณค่าความยาวรันเฉลี่ยโดยวิธีสมการปริพันธ์เชิงตัวเลข (Numerical Integral Equation) ของแผนภูมิควบคุมค่าเฉลี่ยเคลื่อนที่ถ่วงน้ำหนักเลขชี้กำลังดัดแปลง (Modified EWMA) และหลังจากพิสูจน์ทฤษฎีแล้ว 2) พัฒนาโปรแกรมเพื่อจำลองสถานการณ์ในการเปรียบเทียบประสิทธิภาพของแผนภูมิควบคุมคุมค่าเฉลี่ยเคลื่อนที่ถ่วงน้ำหนักเลขชี้กำลังดัดแปลง (Modified EWMA) กับแผนภูมิควบคุมค่าเฉลี่ยเคลื่อนที่ถ่วงน้ำหนักเลขชี้กำลัง(EWMA) ด้วยวิธีวิเคราะห์สูตรสำเร็จ
ผลการวิจัยพบว่า แผนภูมิควบคุมคุมค่าเฉลี่ยเคลื่อนที่ถ่วงน้ำหนักเลขชี้กำลังดัดแปลง สามารถตรวจจับการเปลี่ยนแปลงขนาดเล็กได้ดีกว่าแผนภูมิควบคุมคุมค่าเฉลี่ยเคลื่อนที่ถ่วงน้ำหนักเลขชี้กำลังโดยพิจารณาจากค่าดัชนีต่าง ๆ ได้แก่ ARL1, SDRL1, AEQL และค่า PCI ที่เหมาะสม นอกจากนี้โครงสร้างของแผนภูมิแบบด้านเดียว (one-sided chart) ซึ่งออกแบบมาเพื่อตรวจติดตามการเพิ่มขึ้นของระดับมลพิษของปริมาณฝุ่น PM2.5 ยังช่วยสนับสนุนการประยุกต์ใช้ในงานเฝ้าระวังด้านสิ่งแวดล้อมที่ข้อมูลมีความสัมพันธ์กันในตัวเองได้อย่างมีประสิทธิภาพ มีความรวดเร็ว ความแม่นยำในการตรวจจับความผิดปกติของข้อมูล อีกทั้งยังมีข้อได้เปรียบทั้งในด้านการวิเคราะห์และการคำนวณเมื่อเปรียบเทียบกับวิธีการที่มีอยู่เดิม
This research analyzes and forecasts PM2.5 particulate matter concentrations using a time series approach based on the General Autoregressive Moving Average process, ARMA(p,q), under the assumption that the error terms follow an exponential distribution. A Modified Exponentially Weighted Moving Average (Modified EWMA) control chart is proposed and compared with the traditional Exponentially Weighted Moving Average (EWMA) control chart.
The research methodology consists of two main stages. First, relevant literature, textbooks, and previous studies were reviewed to support the theoretical development. This stage involved deriving an explicit formula for the Average Run Length (ARL) from the integral equation of the Modified EWMA control chart. In addition, the ARL was approximated using the Numerical Integral Equation (NIE) method for the Modified EWMA control chart. After the theoretical derivation was completed, the second stage involved developing a computer program to conduct simulation experiments in order to compare the performance of the Modified EWMA control chart with the traditional EWMA control chart using the explicit formula approach.
The results indicate that the Modified EWMA control chart is more effective at detecting small shifts than the traditional EWMA control chart, as evidenced by several performance measures including ARL1, SDRL1, AEQL, and the optimal Performance Comparison Index (PCI). Furthermore, the one-sided chart structure, which is designed to monitor increases in PM2.5 pollution levels, supports its practical application in environmental monitoring systems where the data exhibit autocorrelation. The proposed method provides fast and accurate detection of abnormal changes in the data and offers advantages in both analytical and computational efficiency compared with existing methods.
|
| คำสำคัญ |
PM2.5,Forecasting,Control chart,Average run length,การพยากรณ์,แผนภูมิควบคุม,ความยาวรันเฉลี่ย
|