Firewall are the very important element in Network security. It perform the task of Surveillance along with sorting Network traffic in order minimize the Unauthorized access as well as malicious activities. Traditional firewall rely on Rule based approaches which may unable to detect all complex and evolving threats effectively. This research propose the framework to improve the effectiveness of firewall anomaly detection using different deep learning algorithm. The proposed Novelty architecture introduced the hybrid deep learning model called Convolution Recurrent Neural network that integrates Artificial Neural Network (ANN), Recurrent neural networks (RNNs), feed forward neural networks, and convolutional neural networks (CNNs). A collection of many perceptron neurons at each level makes up an artificial neural network (ANN). Three layers make up an ANN: input, hidden, and output. Inputs are received by the input layer, processed by the hidden layer, and then produced by the output layer. In essence, every layer attempts to learn certain weights. CNNs are accustomed to capture spatial features from network traffic data, capturing patterns and anomalies in different network protocols and traffic types. Recurrent Neural Network (RNN) is used for pattern recognition for sequential data and speech data analysis. An Inverse Neural Network (FNN) is fundamental type artificial neural network where connections between nodes form a directed cyclic graph without cycles. Architecture will analyses network traffic patterns for identify unusual or suspicious activity and take appropriate action to block or mitigate the threat.