Taguchi Methods were founded by Dr. Genichi Taguchi in the 1950s, and its core content was regarded as a "national treasure" by Japan. This method is suitable for product design, scientific experiments, technological innovation, process reform, material development and other development fields, as well as quality evaluation and quality improvement.

In the early 1960s, Taguchi method won the Deming Award for Quality Application. After the 1980s, AT&T, Ford, Xerox, Motorola, Kodak and other companies successively adopted it. The Taguchi method is an orthogonal experimental technology developed by researchers led by Genichi Taguchi, the Japan Telecommunications Research Institute, based on the experimental design of the Fischer multi-configuration method. It links quality management with economic benefits and integrates engineering experience and statistical principles. , using mathematical methods to comprehensively study quality management theories and methods from multiple perspectives, forming a quality robust design method.

Taguchi theory believes that there is a nonlinear relationship between product target characteristics and the combination of influence parameter levels. Using this characteristic, even if a third-level product component with large characteristic fluctuations is used, relatively stable quality characteristics can be obtained. Japanese technicians often use this method to improve products and production processes, reduce parts and assembly tolerances, and reduce production costs. It is said that 80% of Japan's quality improvement benefits are due to the Taguchi method. In addition, the United States has set up a research association to promote this method, and the former Ministry of Ordnance Industry of my country has also vigorously promoted it, and has achieved certain results in weapon development and functional improvement.

Taguchi method is a robust optimization design method. It uses orthogonal tables to arrange test plans, simulate various interferences that cause product quality fluctuations with error factors, and uses signal-to-noise ratio as an indicator to measure product quality robustness. By conducting statistical analysis of different experimental plans, we can find design plans with strong anti-interference ability, good adjustment ability, stable and reliable performance, and reasonably determine the parameter tolerance based on the principle of minimum quality loss, so as to achieve the lowest cost and best quality technical and economic integration with the lowest quality Effect.

Dr. Taguchi divides product design into three stages: system design, parameter design and tolerance design, also known as three-time design:

System design: When choosing the entire system or configuration for the product, you need to find a variety of different systems, use complete engineering knowledge to evaluate, and choose a design that can meet customer needs with the best technology and the lowest cost.

Parameter design: Find out the main variables that affect product variation in the production process, select a set of parameter standards, and ensure stable product performance. In this stage, the nonlinear effect was used, and the signal-to-noise ratio (SN ratio) was used as the stability index to perform statistical analysis through the orthogonal test method (internal orthogonal and lateral orthogonal). Usually, cheap original components with wide fluctuation range were designed, taking into account both the Improved product quality and cost.

Tolerance design: Based on parameter design, comprehensively consider product quality and production and maintenance costs, determine the optimal tolerance range of various parameters during product production, and use mass loss function and orthogonal polynomial regression analysis for reimprove or redesign. Achieve the best economic benefits.

The Taguchi method aims to make the designed product quality stable and volatile, making the production process insensitive to various noises. In the product design process, high-quality products are developed under low cost conditions by using the functional relationship of quality, cost and efficiency. Taguchi method believes that the benefits of product development can be measured from both internal benefits and social losses in the enterprise. Internal benefits of the enterprise are reflected in low costs under the same functions, while social benefits are measured by the impact of products on people after entering the consumer field. If the product's function fluctuates from its ideal goal and causes losses to society, it means that its robust design is poor, and the Taguchi-style robust design can play a role in reducing costs and reducing product fluctuations.

The mass loss function is the basis of Taguchi's method, which enables decision makers to recognize customer satisfaction, preferences and expectations to determine quality loss, and promotes engineering and technicians to analyze product design and manufacturing processes simultaneously from both technical and economic aspects.

Three designs are three stages that have been experienced in product development or quality improvement, namely system design, parameter design, and tolerance design. In fact, three optimization designs are carried out. The last two designs mainly use orthogonal optimization design to explore product quality and cost. Potential, pursue the effect of less investment, high efficiency and great profit.

System design: that is, functional design, select the basic model system based on the product functions of the market planning, determine the goals and tolerances of product characteristics, and ensure that the product functions meet the standards. In this stage, professional technical theories and methods are mainly used to determine the functions and structure of the system, including the selection of raw materials, parts, components and assembly systems.

Parameter design: It is the core link of product design. The system model is improved and redesigned according to actual production and product usage, and the optimal level combination of all parameters in the system (such as raw materials, parts, components, and components, etc.) is selected to enhance product resistance. Interference, stabilize performance indicators near the target value. Its essence is to use nonlinear effects, use the orthogonal test method of statistical mathematics, and use the signal-to-noise ratio as the stability index of the product quality characteristic value for statistical analysis. At this stage, cheap original components with wide fluctuations are generally designed to achieve dual improvements in quality and cost.

Tolerance design: Based on parameter design, the optimal tolerance range is determined for various parameters during product production. Taking into account product quality and production and maintenance costs, reimprove or redesign the product to achieve the best economic benefits. Tolerance design mainly uses mass loss function and orthogonal polynomial regression analysis.

It can be seen that in the three designs, the system design mainly relies on professional technology, and parameter design and tolerance design belong to the general technical category, relying on mathematics and statistical knowledge.

Select the quality characteristics.

Ideal function for determining quality characteristics.

List all factors that affect this quality characteristic.

Determine the level of the signal factor.

Determine the control factors and determine their levels.

Determine the interference factors and determine their levels, and conduct interference experiments if necessary.

Select the appropriate direct cross-section and arrange a complete experimental plan.

Perform experiments and record experimental data.

Data analysis.

Confirm the experiment. If necessary, the above steps can be repeated until optimal quality and performance are achieved.

It is best to make the error factor compounded to 1 - 2, up to three.

The error factor can be determined based on its influence degree and experimental purpose.

Generally, 2 levels are selected.

5.1.1. Conceptual meaning

5.1.2. Analytical Applications

5.1.3. Comparison of classical methods, Taguchi method and Xie Ning method