Prediction Model of Blasting Fragmentation Based on ACO-BP Neural Network and Its Application

doi:10.12068/j.issn.1005-3026.2025.12.20240122

Abstract

Abstract:

Blasting fragmentation is governed by the combined interplay of rock properties， blasting parameters， and explosive characteristics. Its accurate prediction is the key to achieving the coordinated optimization of precision blasting in mines and energy consumption during mining and processing. To improve prediction accuracy， a blasting fragmentation prediction model was constructed using ant colony optimization （ACO） and a back propagation （BP） neural network， and the key influencing factors of blasting fragmentation were identified. Based on the Sijiaying open pit mine， a blasting case database was established to train the model and improve prediction accuracy， and the weights of the influencing factors were analyzed. Results show that ACO significantly improves model performance. Among the influencing factors of blasting fragmentation， blasthole spacing has the highest weight， and minimum burden has the lowest weight. Blasting parameters exhibit optimal ranges for blasting fragmentation， but single-parameter adjustments cannot sustainably improve blasting fragmentation. This model provides an effective means and theoretical basis for inverse optimization of blasting design based on blasting effect requirements.

Key words: ACO-BP neural network, precision blasting, blasting fragmentation, weight of influencing factor, inverse optimization

CLC Number:

TD 235

Qing-lei YU, Jia-wei WU, You LI, Jiang-yong PU. Prediction Model of Blasting Fragmentation Based on ACO-BP Neural Network and Its Application[J]. Journal of Northeastern University(Natural Science), 2025, 46(12): 116-123.

Figures/Tables 10

Fig.1 Flow chart of ACO-BP algorithm

Fig.2 Acquisition mode of model input parameters

Table 1 Sample data of 11 groups from Sijiaying

样本	R_m	R_c	E	J_V	ρ	S/H	B/H	q	W/H	t/ms	v	ρ_d	d/cm
样本	MPa	MPa	GPa	条⋅m^-3	kg⋅m^-3	S/H	B/H	kg⋅t^-1	W/H	t/ms	m⋅s^-1	g⋅cm^-3	d/cm
1	16.88	175.57	72.53	8.06	3.32	0.57	0.43	0.27	0.29	42	4 500	1.2	25.25
2	17.44	177.90	74.16	7.33	2.69	0.28	0.17	0.24	0.11	42	4 500	1.2	39.52
3	17.44	177.90	74.16	8.43	2.69	0.28	0.17	0.30	0.11	42	4 500	1.2	36.22
4	18.44	198.64	65.46	3.62	2.72	0.21	0.16	0.24	0.11	42	4 500	1.2	34.31
5	15.42	192.46	73.84	3.43	2.74	0.33	0.28	0.25	0.11	42	4 500	1.2	45.82
6	11.33	155.33	55.35	1.84	2.64	0.40	0.30	0.30	0.20	42	4 500	1.2	22.02
7	13.43	183.05	65.37	6.62	2.67	0.27	0.20	0.38	0.13	42	4 500	1.2	30.79
8	17.01	172.02	70.45	5.29	2.65	0.41	0.35	0.16	0.12	42	4 000	1.0	16.11
9	15.37	196.53	66.61	3.33	2.62	0.33	0.28	0.25	0.11	42	4 000	1.0	27.71
10	20.87	218.36	75.66	5.82	2.69	0.47	0.35	0.30	0.12	42	4 500	1.2	19.55
11	12.54	169.36	56.33	3.86	3.38	0.47	0.35	0.29	0.12	42	4 000	1.0	36.18

Table 2 Prediction results and actual values of

测试数据编号	d/cm
测试数据编号	BP神经网络预测值	ACO-BP神经网络预测值	实际值
司家营1#	26.967 0	21.948 1	19.550 0
司家营2#	13.762 1	27.384 9	36.180 0
A矿山1#	33.742 7	28.935 4	29.000 0
A矿山2#	51.043 7	31.058 9	29.700 0
A矿山3#	29.873 5	29.831 2	29.300 0
B矿山1#	20.396 5	23.143 3	35.410 0
B矿山2#	24.891 9	23.876 9	22.840 0
B矿山3#	27.574 3	27.797 5	31.940 0
B矿山4#	27.774 7	24.536 2	21.790 0

Table 3 Index values of prediction results of different

评价指标	BP神经网络预测	ACO-BP神经网络预测
$e R M S$	12.112 6	5.390 2
$e M A P$ /%	31.399 2	11.993 5

Table 3 Index values of prediction results of different

评价指标	BP神经网络预测	ACO-BP神经网络预测
$e R M S$	12.112 6	5.390 2
$e M A P$ /%	31.399 2	11.993 5

Fig.3 Prediction error of model under different training samples

Table 4 Weights from input layer to hidden layer

节点	w_i
节点	R_m	R_c	E	J_V	ρ	S/H	B/H	q	W/H	t	v	ρ_d
1	0.63	0.63	-5.36	-2.75	1.13	-1.59	-0.02	-0.27	-0.56	0.46	-1.72	-0.61
2	1.43	-0.47	2.54	5.27	1.88	0.97	-2.83	0.83	1.60	-2.88	-2.25	-6.90
3	2.25	0.07	-0.16	-2.99	1.50	4.96	0.28	1.28	-1.28	-1.76	-3.96	0.25
4	6.64	1.38	-1.14	-2.74	6.14	6.31	2.45	-3.82	1.59	-3.43	-2.75	-0.31
5	1.01	-0.36	-2.41	6.10	-3.44	4.21	5.57	2.05	0.11	0.53	-0.72	-1.09
6	0.53	3.26	0.14	-2.29	-2.15	-4.81	2.59	4.25	-0.42	0.13	-2.14	-0.86
7	0.32	0.88	-0.83	-2.42	-0.62	7.10	2.41	-5.22	0.88	2.52	0.64	-3.99

Table 5 Weights from hidden layer to output layer

隐含层节点	1	2	3	4	5	6	7
权值	0.409 6	0.060 0	0.143 6	-0.470 8	0.095 2	-0.355 3	-0.302 0

Table 6 Weights of each input parameter in prediction model

参数	R_m	R_c	E	J_V	ρ	S/H	B/H	q	W/H	t	v	ρ_d
权重	0.34	0.24	0.20	-0.28	0.19	0.37	0.32	-0.28	0.15	-0.08	-0.05	-0.12

Fig.4 Influence of main parameters on blasting fragmentation

References 16

[1]	Kulatilake P H S W， Wu Q， Hudaverdi T， et al. Mean particle size prediction in rock blast fragmentation using neural networks［J］. Engineering Geology， 2010， 114（3/4）： 298-311.
[2]	Shi X Z， Zhou J， Wu B B， et al. Support vector machines approach to mean particle size of rock fragmentation due to bench blasting prediction［J］. Transactions of Nonferrous Metals Society of China， 2012， 22（2）：432-441.
[3]	Ke B， Pan R H， Zhang J， et al. Parameter optimization and fragmentation prediction of fan-shaped deep hole blasting in Sanxin Gold and Copper Mine［J］. Minerals， 2022， 12（7）： 788.
[4]	Wang Y C， Guo Q P， Yang S J， et al. A prediction model for blasted block size grouping based on HC and RF-GA-BP neural network［J］. Arabian Journal of Geosciences， 2022， 15 （16）： 1391.
[5]	Huang G B， Zhu Q Y， Siew C K. Extreme learning machine： theory and applications［J］. Neurocomputing， 2006， 70（1/2/3）： 489-501.
[6]	Gao W， Karbasi M， Hasanipanah M， et al. Developing GPR model for forecasting the rock fragmentation in surface mines［J］. Engineering with Computers， 2018， 34（2）： 339-345.
[7]	金长宇，于佳强，王强，等. 基于集成学习CatBoost优化模型的爆堆大块率预测［J］. 东北大学学报（自然科学版）， 2023， 44（12）： 1743-1750.
	Jin Chang-yu， Yu Jia-qiang， Wang Qiang， et al. Prediction of blasting fragment large block percentage ratio based on ensemble learning CatBoost model［J］. Journal of Northeastern University（ Natural Science）， 2023， 44（12）： 1743-1750.
[8]	Gao H， Fu Z L. Forecast of blasting fragmentation distribution based on BP neural network［J］. Advanced Materials Research， 2012， 619： 3-8.
[9]	史秀志，郭霆，尚雪义，等. 基于PCA-BP神经网络的岩石爆破平均粒径预测［J］. 爆破， 2016， 33（2）： 55-61.
	Shi Xiu-zhi， Guo Ting， Shang Xue-yi， et al. Prediction of mean particle size of rock blast based on combination of PCA and BP neural networks［J］. Blasting， 2016， 33（2）： 55-61.
[10]	Guo Q P， Yang S J， Wang Y C， et al. Prediction research for blasting peak particle velocity based on random GA-BP network group［J］. Arabian Journal of Geosciences， 2022， 15（15）： 1351.
[11]	Yu J Y， Ren S J. Prediction and analysis method of mine blasting quality based on GA-BP neural network［J］. Mobile Information Systems， 2022 （1）： 9239381.
[12]	Xing Y H， Li F F. Research on the influence of hidden layers on the prediction accuracy of GA-BP neural network［J］. Journal of Physics： Conference Series， 2020， 1486（2）： 022010.
[13]	Dorigo M， Maniezzo V， Colorni A. Ant system： optimization by a colony of cooperating agents［J］. IEEE Transactions on Systems， Man， and Cybernetics， Part B. Cybernetics，1996， 26（1）： 29-41.
[14]	Li X D， Liu K W， Zhao X R， et al. Study on rock fracturing in smooth blasting under initial stress［J］. Engineering Fracture Mechanics， 2024， 296： 109865.
[15]	赵翔，郭小平，朴志友，等. BP神经网络在爆破块度预测中的应用研究［J］. 水泥技术， 2015 （1）： 36-39.
	Zhao Xiang， Guo Xiao-ping， Zhi-you Piao， et al. Application research of BP neural network in predication of rock fragmentation［J］. Cement Technology， 2015 （1）： 36-39.
[16]	邓飞，肖伟，程秋亭，等. 基于BP神经网络的爆破参数优化［J］. 矿业研究与开发， 2016， 36（4）： 19-21.
	Deng Fei， Xiao Wei， Cheng Qiu-ting， et al. Optimization of blasting parameters based on BP neural network［J］. Mining Research and Development， 2016， 36（4）： 19-21.