Constructing Exponential Models and Real-World Applications — 指数模型的构建与实际应用

📚 Constructing Exponential Models and Real-World Applications | 指数模型的构建与实际应用

指数模型是 A-Level 数学中连接代数与真实世界的核心工具。从人口增长到放射性衰变，从银行复利到咖啡冷却，几乎每一次”数量按固定比例变化”的现象都可以用 y = ab^x 来描述。这篇文章将系统讲解指数模型的构建方法：如何从题目情景写出模型、如何从数据表求出参数、如何用对数把曲线化为直线，以及四大经典应用场景的完整解题过程。全文中英双语对照，适合 Edexcel、AQA、OCR 等各考试局 A-Level 数学考生复习使用。

Exponential models are the core tool in A-Level Mathematics that connects algebra with the real world. . From population growth to radioactive decay, from bank compound interest to a cooling cup of coffee, almost every phenomenon in which “a quantity changes by a fixed ratio” can be described by y = ab^x. This article systematically explains how to construct exponential models: how to write a model from a given scenario, how to find the parameters from a data table, how to use logarithms to turn a curve into a straight line, and the complete solution processes for four classic applications. The article is fully bilingual with Chinese and English paired throughout, and is suitable for students preparing for A-Level Mathematics under Edexcel, AQA, OCR and other exam boards.

1. What Is an Exponential Model? Form, Parameters and Meaning | 什么是指数模型：形式、参数与含义

指数模型的标准形式是 y = ab^x，其中 y 是因变量，x 是自变量（通常代表时间），a 是初始值（即 x = 0 时 y 的取值），b 是变化因子。与线性模型 y = mx + c 不同，指数模型的变化率本身也在变化：每经过一个固定的时间间隔，y 的值都乘以同一个倍数 b。例如 y = 3 × 2^x 在 x = 0, 1, 2, 3 时分别取 3, 6, 12, 24，每一步都翻一倍，这就是”按比例变化”的含义。

The standard form of an exponential model is y = ab^x d time interval, the value of y is multiplied by the same factor b. For example, y = 3 × 2^x values 3, 6, 12, 24 at x = 0, 1, 2, 3, doubling at every step, and that is exactly what “changing by a fixed ratio” means.

为什么指数模型如此重要？因为在真实世界中，很多数量的变化率与它当前的大小成正比：人口越多，每年新增的人口越多；放射性原子越多，每秒衰变的原子越多；存款越多，每年产生的利息越多。这种”自我放大”的机制用线性模型无法描述，只有指数模型能够准确刻画。这也是为什么 A-Level 考纲把”指数与对数”单列一章，并且几乎所有考试局都会出 1 到 2 道相关的应用题。

Why are exponential models so important? Because in the real world, the rate of change of many quantities is proportional to their current size: the larger the population, the more new people are added each year; the more radioactive atoms, the more atoms decay per second; the more money in a deposit, the more interest it earns each year. This “self-amplifying” mechanism cannot be described by a linear model; only an exponential model captures it accurately. This is why the A-Level specification gives “Exponentials and Logarithms” its own chapter, and why almost every exam board sets one or two application questions on it.

2. Exponential Growth vs Exponential Decay: Reading the Base | 指数增长与指数衰减：从底数判断变化方向

在 y = ab^x 中，底数 b 的大小决定了变化方向。如果 b > 1，函数随 x 增大而增大，称为指数增长；如果 0 < b < 1，函数随 x 增大而减小，称为指数衰减。注意 a 必须大于 0，因为实际情景中的数量（人口、质量、金额）不可能是负数。举个例子：y = 100 × 1.05^x 表示初始值 100、每单位时间增长 5% 的模型；而 y = 100 × 0.95^x 表示初始值 100、每单位时间减少 5% 的模型。

In y = ab^x n real scenarios (population, mass, money) cannot be negative. For example, y = 100 × 1.05^x represents a model with initial value 100 that grows by 5% per unit time, while y = 100 × 0.95^x represents a model with initial value 100 that decreases by 5% per unit time.

有一个高频易错点：增长率与增长因子的区别。如果说”每年增长 5%”，那么 b = 1 + 5% = 1.05；如果说”每年减少 5%”，那么 b = 1 − 5% = 0.95。很多同学看到 5% 就直接写 b = 0.05 或 b = 5，这是对”变化因子”理解不到位。考试中常把增长率记为 r，模型写作 y = a(1 + r)^x，其中 r 为正表示增长、为负表示衰减。判断 b 与 1 的大小关系是这类题的第一步，也是最容易拿分的一步。

fall: the difference between the growth rate and the growth factor. If the question says “grows by 5% per year”, then b = 1 + 5% = 1.05; if it says “decreases by 5% per year”, then b = 1 − 5% = 0.95. Many students see 5% and immediately write b = 0.05 or b = 5, which shows a weak grasp of the “change factor&# . In exams the growth rate is often written as r, and the model is written as y = a(1 + r)^x, where a positive r means growth and a negative r means decay. Comparing b with 1 is the first step of this type of question, and also the easiest mark to secure.

3. Building a Model from a Data Table: The Constant-Ratio Test | 从数据表构建模型：常数比值检验

构建指数模型的第一步，是判断数据是否真的服从指数规律。标准方法是”常数比值检验”：如果 x 等间隔取值，而相邻 y 值的比值 y₂/y₁、y₃/y₂、y₄/y₃ 都近似相等，那么数据就符合指数模型。例如某实验测得：x = 0, 1, 2, 3 时 y = 4.0, 8.1, 16.2, 32.4，相邻比值依次为 8.1/4.0 = 2.025、16.2/8.1 = 2.000、32.4/16.2 = 2.000，非常接近常数 2，说明 y 大致服从 y = 4 × 2^x。

to judge whether the data really follows an exponential law. The standard method is the “constant-ratio test&# and 32.4/16.2 = 2.000, very close to the constant 2, so y approximately follows y = 4 × 2^x.

与之相对，如果相邻差值 y₂ − y₁、y₃ − y₂、y₄ − y₃ 近似相等，则数据更符合线性模型 y = mx + c。这一检验在考试中经常以”解释为什么数据符合指数模型”的形式出现，占 2 到 3 分。答题时一定要写清楚两步：第一，相邻 y 值的比值近似恒定；第二，这说明 y 按固定倍数变化，因此可用指数模型描述。只写”看起来是指数的”不给分。

ual, the data fits a linear model y = mx + c better. This test often appears in exams as “explain why the data follows an exponential model&# ond, this means y changes by a fixed factor, so the data can be described by an exponential model. Writing only “it looks exponential” earns no marks.

4. Log-Linearisation: Turning y = ab^x into a Straight Line | 对数线性化：把 y = ab^x 化为直线

构建指数模型的核心技巧是对数线性化。对 y = ab^x 两边取自然对数，利用对数运算律得到 ln y = ln a + x ln b。如果把 ln y 看作新的纵坐标、x 看作横坐标，这就是一条斜率为 ln b、截距为 ln a 的直线。因此，当题目给出一组 (x, y) 数据并要求构建模型时，标准做法是：先计算每一行的 ln y，再对 (x, ln y) 这些点拟合直线（手算或用计算器的线性回归功能），读出斜率与截距，最后还原 a = e^截距、b = e^斜率。

exponential model is log-linearisation. Taking natural logarithms of both sides of y = ab^x ard procedure is: first compute ln y for each row, then fit a straight line to the (x, ln y) points (by hand or with the linear regression function of a calculator), read off the gradient and intercept, and finally recover a = e^intercept and b = e^gradient.

注意两个细节。第一，两边取对数时必须用同一个底数，通常取自然对数 ln（也有题目用 log₁₀，这时还原公式变成 a = 10^截距、b = 10^斜率，务必看清底数）。第二，如果题目直接给出最佳拟合直线的方程，比如 ln y = 2.1 + 0.35x，那么立刻得到 ln a = 2.1、ln b = 0.35，从而 a = e².1 ≈ 8.17，b = e⁰.35 ≈ 1.42，模型就是 y = 8.17 × 1.42^x。这类题考的是”对数运算律”与”直线方程”两个知识点的结合，属于必拿分题型。

logarithm ln (some questions use log10, in which case the recovery formulas become a = 10^intercept and b = 10^gradient fit, for example ln y = 2.1 + 0.35x, then immediately ln a = 2.1 and ln b = 0.35, so a = e².1 ≈ 8.17 and b = e⁰.35 ≈ 1.42, giving the model y = 8.17 × 1.42^x. This type of question tests the combination of the “laws of logarithms” and “straight-line equations”, and is a guaranteed mark if you know the method.

5. Finding a and b from Two Data Points | 由两个数据点求参数 a 和 b

如果题目只给两个数据点 (x₁, y₁) 和 (x₂, y₂)，不需要拟合，直接解方程组即可。把两个点代入 y = ab^x，得到 y₁ = ab^x₁ 和 y₂ = ab^x₂。两式相除消去 a，得到 y₂/y₁ = b^{x₂ − x₁}，从而 b = (y₂/y₁)^(1/(x₂ − x₁))；再把 b 代回任意一式求出 a。例如点 (0, 5) 与 (4, 80)：b = (80/5)^1/4 = 16^1/4 = 2，a = 5，模型为 y = 5 × 2^x。

is no need to fit; simply solve a pair of equations. Substituting both points into y = ab^x gives y1 = ab^x1 and y2 = ab^x2. Dividing the two equations eliminates a, giving y2/y1 = b^{x2 − x1} ck into either equation to find a. For example, with points (0, 5) and (4, 80): b = (80/5)^1/4 = 16^1/4 = 2, a = 5, so the model is y = 5 × 2^x.

这个方法的优点是计算量小，缺点是只用了两个点，对测量误差非常敏感。考试中通常在题目里明确说明”模型经过这两个点”或者给出”初始值与某时刻的值”，此时直接代入即可。注意：x₂ − x₁ 出现在指数位置上，如果两个点的横坐标间隔不是整数，就要用分数指数或对数来求 b，例如 b = e^(ln(y₂/y₁)/(x₂ − x₁))。另外，如果第一个点的横坐标不是 0，a 就不再等于 y₁，必须用完整方程组求解，这是很多同学容易忽略的地方。

ery sensitive to measurement error. In exams the question usually states explicitly that “the model passes through these two points” or gives “the initial value and the value at a certain time&# − x1)). Also, if the first point does not have x-coordinate 0, then a is no longer equal to y1 and you must solve the full pair of equations, which many students overlook.

6. Half-Life and Doubling Time: The Two Key Rates | 半衰期与倍增时间：两个关键速率

指数模型有两个重要的特征量：半衰期与倍增时间。半衰期指衰减的数量降到初始值一半所需的时间，记为 T½；倍增时间指增长的数量达到初始值两倍所需的时间，记为 Td。对于模型 y = ab^x，若 x 以年为单位，半衰期满足 ab^T½ = a/2，解得 b^T½ = 1/2，即 T½ = ln(1/2)/ln b = −ln 2/ln b。因为衰减时 ln b < 0，所以 T½ 恒为正数。同理，倍增时间 Td = ln 2/ln b。

for a growing quantity to reach twice its initial value, written Td. For the model y = ab^x, if x is measured in years, the half-life satisfies ab^T1/2 = a/2, which gives b^T1/2 b < 0 for decay, T1/2 is always positive. Similarly, the doubling time is Td = ln 2/ln b.

这两个公式把抽象的底数 b 翻译成直观的语言：”多长时间翻一倍”或”多长时间减一半”。例如某放射性同位素每年衰减 3%，即 b = 0.97，则半衰期 T½ = −ln 2/ln 0.97 ≈ 22.8 年，意思是大约 23 年后剩余质量约为原来的一半。反过来，如果题目告诉你半衰期是 5 年，就可以反求 b：b = (1/2)^1/5 ≈ 0.871，即每年剩余 87.1%。建议把下面两个公式记牢并理解推导过程，因为考试常以”证明”或”解释”的形式考查：

These two formulas translate the abstract base b into plain language: “how long until it doubles” or “how long until it halves”. For example, a radioactive isotope decays by 3% ly, if a question tells you the half-life is 5 years, you can find b in reverse: b = (1/2)^1/5 ≈ 0.871, meaning 87.1% las below and understand their derivations, because exams often test them in the form of “prove” or “explain”:

Quantity 特征量	Formula 公式	Meaning 含义
Half-life 半衰期 T½	T½ = −ln 2 / ln b	Time to fall to half 降到一半所需时间
Doubling time 倍增时间 Td	Td = ln 2 / ln b	Time to grow to double 增长到两倍所需时间

7. Real-World Application: Population Growth | 实际应用：人口增长模型

人口增长是考试中最常见的指数模型情景。设某城市 2020 年人口为 80 万，之后每年以 2.4% 的速度增长，则模型为 P(t) = 800000 × 1.024^t，其中 t 为从 2020 年起经过的年数。要求 2035 年的人口，代入 t = 15：P(15) = 800000 × 1.024¹⁵ ≈ 800000 × 1.427 ≈ 1,142,000，约 114 万。要求人口翻倍的时间，用倍增时间公式 Td = ln 2/ln 1.024 ≈ 29.2 年，即大约在 2049 年人口达到 160 万。

cenario in exams. Suppose a city had a population of 800,000 in 2020 and then grows at 2.4% per year; the model is P(t) = 800000 × 1.024^t ears since 2020. To find the population in 2035, substitute t = 15: P(15) = 800000 × 1.024¹⁵ a Td = ln 2/ln 1.024 ≈ 29.2 years, meaning the population reaches 1.6 million around 2049.

广告 / Sponsored

答题时要注意单位的统一：如果增长率是”每年 2.4%”，时间 t 就必须以年为单位；如果题目说”每 10 年翻一番”，那么 x 每增加 1 代表 10 年，此时底数 b = 2，模型写作 P(t) = P₀ × 2^t/10。这类题目经常附带一个追问：”解释为什么该模型不可能长期成立”。标准答案是：现实人口受资源、土地、政策、疾病等因素限制，增长率会随人口规模变化，不可能无限指数增长，因此模型只在有限时间范围内有效。

When answering, make sure the units are consistent: if the growth rate is “2.4% per year”, then the time t must be measured in years; if the question says “doubles every 10 years&# of 1 in x represents 10 years, so the base is b = 2 and the model is written P(t) = P0 × 2^t/10. These questions often include a follow-up: “explain why this model cannot hold in the long term”. The standard answer is that real populations are limited by resources, land, policy, disease and other factors, and the growth rate changes with the size of the population, so unlimited exponential growth is impossible; the model is only valid within a finite time range.

8. Real-World Application: Radioactive Decay and Carbon Dating | 实际应用：放射性衰变与碳定年

放射性衰变是典型的指数衰减模型。设某样品初始质量为 m₀，衰变常数为 k（每秒剩余的比例为 e^−k），则 t 秒后的质量 m(t) = m₀e^−kt。写成 e 的幂是为了方便后续求导和积分。题目常给半衰期：例如碳-14 的半衰期约为 5730 年，那么 k = ln 2/5730 ≈ 1.21 × 10⁻⁴（每年），模型为 m(t) = m₀e^{−1.21×10⁻⁴ t}。

Radioactive decay is the classic exponential decay model. Suppose a sample has initial mass m0 and decay constant k (the fraction remaining after each second is e^−k); then the mass after t seconds is m(t) = m0e^−kt : for example, carbon-14 has a half-life of about 5730 years, so k = ln 2/5730 ≈ 1.21 × 10⁻⁴ per year, and the model is m(t) = m0e^(−1.21×10⁻⁴t).

碳定年法是经典考题：考古发现一件木制文物，测得其中碳-14 的剩余量是原来的 45%，问文物的年代。设年代为 t 年，则 0.45 = e^−kt，两边取自然对数得 −kt = ln 0.45，解得 t = −ln 0.45/k = −ln 0.45 × 5730/ln 2 ≈ 6580 年。解题的关键两步是：第一，写出”剩余比例 = e^−kt”这个关系；第二，把半衰期换算成衰变常数 k = ln 2/T½。这两个步骤各占 2 到 3 分，缺一不可。

dating is a classic exam question: an archaeological wooden artefact is found to retain 45% s original carbon-14, and you are asked for its age. Let the age be t years; then 0.45 = e^−kt −ln 0.45 × 5730/ln 2 ≈ 6580 years. The two key steps are: first, write down the relation “remaining fraction = e^−kt&# k = ln 2/T1/2. Each of these two steps is worth 2 to 3 marks, and neither can be omitted.

9. Real-World Application: Compound Interest and Depreciation | 实际应用：复利与折旧

金融中的复利计算本质上就是指数模型。设本金为 P，年利率为 r（写成小数），每年复利一次，则 n 年后本息和 A = P(1 + r)ⁿ。例如本金 10,000 元，年利率 4%，10 年后 A = 10000 × 1.04¹⁰ ≈ 14,802 元。如果每半年复利一次，则每期利率为 r/2、期数为 2n，公式变为 A = P(1 + r/2)²ⁿ；每季度复利则 A = P(1 + r/4)⁴ⁿ。复利频率越高，相同年利率下的最终金额越大。

interest is compounded once per year; then after n years the total amount is A = P(1 + r)ⁿ. For example, with principal 10,000 yuan and an annual rate of 4%, after 10 years A = 10000 × 1.04¹⁰ e per period is r/2 and the number of periods is 2n, so the formula becomes A = P(1 + r/2)²ⁿ; compounding quarterly gives A = P(1 + r/4)⁴ⁿ. The more frequently interest is compounded, the larger the final amount for the same annual rate.

折旧则是复利的镜像：资产价值按固定百分比逐年下降，模型为 V = V₀(1 − d)ⁿ，其中 d 是年折旧率。例如一辆车购入价 20 万元，每年贬值 15%，则 5 年后价值 V = 200000 × 0.85⁵ ≈ 88,737 元。考试中常见追问：”价值降到一半需要多少年？”此时用半衰期公式 n = ln 0.5/ln 0.85 ≈ 4.27 年。另外要注意区分”离散复利”与”连续复利”：当复利频率无限增大时，模型趋于 A = Pe^rn，这是 e 的定义在金融中的一个直接应用。

he value of an asset falls by a fixed percentage each year, giving the model V = V0(1 − d)ⁿ the annual depreciation rate. For example, a car bought for 200,000 yuan depreciates by 15% per year, so after 5 years its value is V = 200000 × 0.85⁵ ≈ 88,737 yuan. A common follow-up question is: “how many years until the value halves?&# se the half-life formula n = ln 0.5/ln 0.85 ≈ 4.27 years. Also be careful to distinguish “discrete compounding” from “continuous compounding”: as the compounding frequency increases without bound, the model tends to A = Pe^rn, which is a direct application of the definition of e in finance.

10. Real-World Application: Newton’s Law of Cooling | 实际应用：牛顿冷却定律

牛顿冷却定律描述物体温度随时间趋于环境温度的过程：物体温度 T 与环境温度 T₀ 的差按指数衰减，即 T(t) − T₀ = (T(0) − T₀)e^−kt。例如一杯 90°C 的咖啡放在 20°C 的房间里，若 10 分钟后温度为 60°C，则温差从 70°C 降到 40°C，代入得 40 = 70e^−10k，所以 e^−10k = 4/7，k = −ln(4/7)/10 ≈ 0.0560（每分钟）。要求咖啡降到 30°C 的时间：温差为 10°C，则 10 = 70e^−kt，即 e^−kt = 1/7，t = ln 7/k ≈ 34.8 分钟。

Newton&# e T and the ambient temperature T0 decays exponentially, that is, T(t) − T0 = (T(0) − T0)e^−kt 0°C after 10 minutes, the temperature difference has fallen from 70°C to 40°C, so 40 = 70e^−10k, giving e^−10k ute. To find when the coffee reaches 30°C: the temperature difference is 10°C, so 10 = 70e^−kt, that is, e^−kt = 1/7, and t = ln 7/k ≈ 34.8 minutes.

这类题的难点在于：模型描述的是”温差”而不是”温度”本身，因此必须先算出 T(t) − T₀ 再代入。另一个常见错误是把环境温度 T₀ 当成 0 处理，直接对 T 用指数模型，结果完全错误。答题建议按三步走：第一步写出温差形式的方程；第二步代入已知点求 k；第三步解目标方程求时间或温度。值得注意，e^−kt 中 k 的单位与时间单位必须匹配，例如 k 是”每分钟”，时间就必须用分钟。Edexcel 近年的真题多次考查这一模型，务必熟练掌握。

The difficulty of these questions is that the model describes the “temperature difference” rather than the temperature itself, so you must first compute T(t) − T0 before substituting. Another common error is treating the ambient temperature T0 as zero and applying an exponential model to T directly, which gives a completely wrong result. A three-step approach is recommended: first write the equation in terms of the temperature difference; second substitute a known point to find k; third solve the target equation for the time or temperature. Note that the units of k in e^−kt must match the time units: if k is “per minute”, then time must be measured in minutes. Real Edexcel papers in recent years have examined this model several times, so you should master it thoroughly.

11. Exam Question Framework: The Five-Step Method | 考试题型：五步解题法

综合以上内容，A-Level 指数模型大题可以总结为五步。第一步，识别情景类型（增长、衰减还是冷却），写出模型的一般形式，并说明每个参数的含义，这一步通常是题目明确要求的 1 到 2 分。第二步，利用题目给出的数据（初始值、某个时刻的值、增长率或半衰期）求出 a 和 b，必要时取对数线性化。第三步，把题目要求的时间或数量代入模型求解。第四步，如果需要比较或判断，取对数把指数方程化为线性方程再解。第五步，检查答案的合理性：增长模型的结果应随时间增大，衰减模型的结果应随时间减小，任何负值都说明计算有误。

Putting everything together, a full A-Level exponential model question can be solved in five steps. Step 1: identify the scenario type (growth, decay or cooling), write down the general form of the model and state the meaning of each parameter; this step is usually worth the explicitly requested 1 to 2 marks. Step 2: use the data given in the question (initial value, value at a certain time, growth rate or half-life) to find a and b, using log-linearisation if necessary. Step 3: substitute the required time or quantity into the model and solve. Step 4: if comparison or judgement is needed, take logarithms to turn the exponential equation into a linear one and solve. Step 5: check that the answer is reasonable: the result of a growth model should increase with time, the result of a decay model should decrease with time, and any negative value indicates a calculation error.

时间分配上，这类题目通常占 6 到 9 分，建议控制在 10 到 15 分钟内完成。写答案时把”模型形式、参数代入、解方程、结论”四步分开写，即使最终数值算错，中间步骤也能拿到方法分。特别提醒：用计算器求 e^x 或 ln 时，注意题目要求保留几位有效数字（通常为 3 s.f.），因为四舍五入的误差在后续代入中会被放大，导致最后一位答案偏差。

In terms of time management, these questions are usually worth 6 to 9 marks, and it is recommended to finish them within 10 to 15 minutes. When writing your answer, separate the four stages “model form, parameter substitution, equation solving, conclusion” so that even if the final number is wrong, the intermediate steps still earn method marks. One special reminder: when using a calculator to find e^x or ln, note how many significant figures the question asks for (usually 3 s.f.), because rounding errors are amplified in later substitutions and can shift the final digit of the answer.

12. Common Mistakes and How to Avoid Them | 常见错误与规避方法

第一个高频错误是把增长率与增长因子混淆：增长 5% 对应 b = 1.05，而不是 b = 0.05 或 b = 5。规避方法：凡是看到百分数变化，先写成&#8221;1 ± 百分数”再判断大小关系。第二个高频错误是对数线性化后忘记还原：从 ln y = ln a + x ln b 得到斜率 m 和截距 c 后，必须用 a = e^c、b = e^m 还原，很多同学直接把 m 当 b、把 c 当 a 代入，答案差了十万八千里。

The first high-frequency error is confusing the growth rate with the growth factor: a 5% = 0.05 or b = 5. The way to avoid it: whenever you see a percentage change, first write &#8220;1 ± percentage&# obtaining the gradient m and intercept c from ln y = ln a + x ln b, you must recover a = e^c and b = e^m, but many students substitute m as b and c as a directly, and the answer is wildly wrong.

第三个错误是忽略定义域：x 代表时间，通常 x ≥ 0，但题目有时会问”模型在 x < 0 时是否合理"。例如人口模型在 t 为负时给出一个很小的正数，这在数学上成立但在现实中没有意义，因为不存在"负的时间"。第四个错误是单位不统一：增长率按年给出、时间却用月，或半衰期用天、时间用年。最后，务必检查半衰期公式的符号：衰减模型中 ln b < 0，公式 T½ = −ln 2/ln b 前面的负号不能丢，否则会得到负数时间。

ignoring the domain: x represents time and is usually x ≥ 0, but questions sometimes ask “is the model reasonable for x < 0" hich is mathematically valid but meaningless in reality, because there is no such thing as "negative time" nus sign in T1/2 = −ln 2/ln b must not be dropped, otherwise you will get a negative time.

Summary | 总结

指数模型 y = ab^x 是 A-Level 数学中连接代数、对数与真实世界的桥梁。本文系统讲解了它的标准形式与参数含义、增长与衰减的判断方法、从数据构建模型的常数比值检验、取对数线性化的核心技巧、由两个数据点求参数、半衰期与倍增时间的计算，以及人口增长、放射性衰变、复利折旧、牛顿冷却四大经典应用。这些知识点在 Edexcel、AQA、OCR 等考试局的试卷中反复出现，是 A-Level 纯数部分的高频考点。

The exponential model y = ab^x is a bridge in A-Level Mathematics connecting algebra, logarithms and the real world. This article has systematically covered its standard form and parameter meanings, judging growth versus decay, the constant-ratio test for building models from data, the core technique of log-linearisation, finding parameters from two data points, half-life and doubling time, and the four classic applications of population growth, radioactive decay, compound interest/depreciation and Newton’s law of cooling. These knowledge points appear repeatedly in the papers of Edexcel, AQA, OCR and other exam boards, and are high-frequency topics in the pure mathematics part of A-Level.

复习建议：把本文的例题独立重做一遍，不要边看答案边做；然后找 3 到 5 道历年真题练习，重点练”从情景写模型”和”取对数线性化”两个环节。掌握五步解题法和四个常见错误的规避方法之后，指数模型类题目就可以稳定拿分，成为你 A-Level 数学考试中的送分题。

Revision advice: redo the worked examples in this article independently, without peeking at the solutions; then practise with 3 to 5 past-paper questions, focusing on the two stages of “writing a model from a scenario” and “log-linearisation”. Once you master the five-step method and the ways to avoid the four common errors, exponential model questions become a reliable source of marks and a gift question in your A-Level Mathematics exam.

更多咨询请联系16621398022（同微信）

更多咨询请联系16621398022（同微信）