煤炭是我国主要的生产能源和国民基础，在我国经济建设、能源供应中发挥着重要作用。在煤矿开采中，煤层中会有瓦斯不断涌出，高效实施瓦斯抽采是降低煤矿瓦斯事故发生率的重要手段。差压式气体流量计在矿用瓦斯抽采管道流量计量中得到广泛应用。然而，瓦斯抽采管道环境的复杂性给差压式流量计的准确测量带来了诸多挑战，导致其测量精度难以满足实际需求。因此需要对差压式流量计进行误差补偿。针对上述问题，本文做了如下研究：

针对瓦斯抽采管道中特殊环境可能会对差压式气体流量计产生较大影响的现状，设计差压式气体流量计测量误差实验台，模拟管道环境，验证在温度、湿度、气压三种因素对差压式气体流量计输出值的影响。通过记录流量计的测量值，得出了其输出值随温度、湿度、气压以及多因素变化的曲线；实验结果表明，同一流量点的测量值与温度、湿度的变化呈正相关，与气压的变化呈负相关，其平均测量误差分别为7.34%，5.07%以及6.11%；同时改变三种影响因素，流量计原始测量误差可达7.4%，证明温度、湿度、气压三种因素对差压式气体流量计影响较大。

针对矿用差压式流量计易受井下瓦斯抽采管道中温度、湿度、压力等因素干扰，导致测量误差较大的问题，提出PSO-LSSVM误差补偿模型对差压式气体流量计进行误差补偿。针对因手动调参导致最小二乘支持向量机（Least Squares Support Vector Machine，LSSVM）难以快速找到最优解的问题，利用粒子群算法（Particle Swarm Optimization, PSO）优化LSSVM的核函数参数和惩罚参数。训练误差补偿模型及对比模型后，实验结果表明，在三种单一影响因素下，PSO-LSSVM算法使最大百分比误差分别从原始误差的4.72%、3.04%和1.95%下降到0.45%、0.09%和0.08%。将补偿结果与LSSVM与BP神经网络对比分析发现，LSSVM使得差压式流量计最大百分比误差分别下降到1.19%、0.19%和0.09%；BP神经网络使最大百分比误差分别下降到1.51%、1.58%和1.54%。对比结果表明在单一因素影响下，PSO-LSSVM算法能够是流量计检测精度更高，补偿效果更好。

针对矿用差压式流量计易受井下瓦斯抽采管道中温度、湿度、压力等多因素共同干扰，导致测量误差较大的问题，提出了一种基于改进的鲸鱼算法（Improved Whale Optimization Algorithm，IWOA）优化LSSVM的差压式流量计误差补偿方法。首先，针对PSO在处理多维问题时，容易陷入局部最优解的问题，采用鲸鱼算法优化LSSVM的核函数参数和惩罚参数。其次，引入自适应权重、随机性学习方法以及Tent混沌映射，改善鲸鱼算法种群分布不均、收敛速度慢等问题，构建IWOA-LSSVM误差补偿模型。采用IWOA-LSSVM方法对矿用差压式气体流量计进行误差补偿，对比PSO-BP、IWOA-BP及PSO-LSSVM算法，结果表明：采用PSO-BP、IWOA-BP及PSO-LSSVM算法，流量计原始误差从7.40%分别下降到1.13%、0.61%和1.05%，采用IWOA-LSSVM算法使百分比误差下降到0.23%。IWOA-LSSVM算法能有效消除环境因素对流量计输出结果的影响，提高了矿用差压式气体流量计的可靠性与检测精度。

本研究工作能够有效补偿差压式气体流量计测量误差，显著改善了其性能，能够更好地适应煤矿瓦斯抽采管道中温度、湿度、压力等多种环境因素的变化，保证了流量检测的准确性，为瓦斯抽采管道流量的精准测量提供技术支持。

Coal serves as the primary production energy and national foundation in China, playing a critical role in the country's economic construction and energy supply. In coal mining operations, gas continuously emanates from coal seams, and the efficient implementation of gas drainage is a vital means to reduce the incidence of coal mine gas accidents. Differential pressure gas flowmeters have been widely applied in the flow measurement of mine gas drainage pipelines. However, the complexity of the gas drainage pipeline environment poses numerous challenges to the accurate measurement of differential pressure flowmeters, making it difficult for their measurement precision to meet practical requirements. Therefore, error compensation for differential pressure flowmeters is necessary. To address the above issues, this thesis conducts the following research:

In light of the potential significant impact of the unique environment in gas extraction pipelines on differential pressure gas flow meters, an experimental platform for measuring errors of differential pressure gas flow meters was established to simulate pipeline conditions. This platform verified the influence of temperature, humidity, and pressure on the output values of differential pressure gas flow meters. By recording the flow meter's measurements, curves depicting the variations in output values with temperature, humidity, pressure, and multiple factors were derived. The experimental results indicated that at the same flow rate, the measured values exhibited a positive correlation with changes in temperature and humidity and a negative correlation with changes in pressure, with average measurement errors of 7.34%, 5.07%, and 6.11%, respectively. When all three factors were varied simultaneously, the original measurement error of the flow meter reached 7.4%, demonstrating the substantial impact of temperature, humidity, and pressure on differential pressure gas flow meters.

To address the issue of mine-used differential pressure flow meters being susceptible to interference from temperature, humidity, and pressure in underground gas extraction pipelines, leading to significant measurement errors, a PSO-LSSVM error compensation model was proposed. To overcome the challenge of manually tuning parameters in the Least Squares Support Vector Machine (LSSVM), which hinders rapid identification of the optimal solution, the Particle Swarm Optimization (PSO) algorithm was employed to optimize the kernel function parameters and penalty parameters of LSSVM. After training the error compensation model and comparative models, the experimental results showed that under the influence of each single factor, the PSO-LSSVM algorithm reduced the maximum percentage errors from the original errors of 4.72%, 3.04%, and 1.95% to 0.45%, 0.09%, and 0.08%, respectively. A comparative analysis with LSSVM and BP neural network revealed that LSSVM reduced the maximum percentage errors of the differential pressure flow meter to 1.19%, 0.19%, and 0.09%, while the BP neural network reduced them to 1.51%, 1.58%, and 1.54%. The comparison demonstrated that under single-factor influence, the PSO-LSSVM algorithm achieved higher detection accuracy and better compensation performance.

To tackle the problem of mine-used differential pressure flow meters being prone to significant measurement errors due to the combined interference of multiple factors such as temperature, humidity, and pressure in underground gas extraction pipelines, an error compensation method based on an Improved Whale Optimization Algorithm (IWOA) optimized LSSVM was proposed. First, to address the tendency of PSO to converge to local optima when handling multidimensional problems, the Whale Optimization Algorithm (WOA) was adopted to optimize the kernel function parameters and penalty parameters of LSSVM. Subsequently, adaptive weighting, stochastic learning methods, and Tent chaotic mapping were introduced to improve issues such as uneven population distribution and slow convergence in WOA, thereby constructing the IWOA-LSSVM error compensation model. The application of the IWOA-LSSVM method for error compensation of mine-used differential pressure gas flow meters, compared with PSO-BP, IWOA-BP, and PSO-LSSVM algorithms, yielded the following results: the original error of the flow meter was reduced from 7.40% to 1.13%, 0.61%, and 1.05% using PSO-BP, IWOA-BP, and PSO-LSSVM algorithms, respectively, while the IWOA-LSSVM algorithm reduced the percentage error to 0.23%. The IWOA-LSSVM algorithm effectively mitigated the impact of environmental factors on the flowmeter's output, enhancing the reliability and measurement accuracy of mine-used differential pressure gas flow meters.

This research work can effectively compensate the measurement errors of differential pressure gas flowmeters, significantly improve their performance, better adapt to the changes in multiple environmental factors such as temperature, humidity, and pressure in coal mine gas drainage pipelines, ensure the accuracy of flow detection, and provide technical support for the precise measurement of gas drainage pipeline flow.